This is an internal, incomplete preview of a proposed change to the ACL Anthology.
For efficiency reasons, we generate only three BibTeX files per volume, and the preview may be incomplete in other ways, or contain mistakes.
Do not treat this content as an official publication.
Natural languages are believed to be (mildly) context-sensitive. Despite underpinning remarkably capable large language models, transformers are unable to model many context-free language tasks. In an attempt to address this limitation in the modeling power of transformer-based language models, we propose augmenting them with a differentiable, stack-based attention mechanism. Our stack-basedattention mechanism can be incorporated into any transformer-based language model and adds a level of interpretability to the model. We show that the addition of our stack-based attention mechanism enables the transformer to model some, but not all, deterministic context-freelanguages.
The recent successes and spread of large neural language models (LMs) call for a thorough understanding of their abilities. Describing their abilities through LMs’ representational capacity is a lively area of research. Investigations of the representational capacity of neural LMs have predominantly focused on their ability to recognize formal languages. For example, recurrent neural networks (RNNs) as classifiers are tightly linked to regular languages, i.e., languages defined by finite-state automata (FSAs). Such results, however, fall short of describing the capabilities of RNN language models (LMs), which are definitionally distributions over strings. We take a fresh look at the represen- tational capacity of RNN LMs by connecting them to probabilistic FSAs and demonstrate that RNN LMs with linearly bounded precision can express arbitrary regular LMs.
Plenty of existing work has analyzed the abilities of the transformer architecture by describing its representational capacity with formal models of computation. However, the focus so far has been on analyzing the architecture in terms of language acceptance. We contend that this is an ill-suited problem in the study of language models (LMs), which are definitionally probability distributions over strings. In this paper, we focus on the relationship between transformer LMs and n-gram LMs, a simple and historically relevant class of language models. We show that transformer LMs using the hard or sparse attention mechanisms can exactly represent any n-gram LM, giving us a concrete lower bound on their probabilistic representational capacity. This provides a first step towards understanding the mechanisms that transformer LMs can use to represent probability distributions over strings.
For nearly three decades, language models derived from the n-gram assumption held the state of the art on the task. The key to their success lay in the application of various smoothing techniques that served to combat overfitting. However, when neural language models toppled n-gram models as the best performers, n-gram smoothing techniques became less relevant. Indeed, it would hardly be an understatement to suggest that the line of inquiry into n-gram smoothing techniques became dormant. This paper re-opens the role classical n-gram smoothing techniques may play in the age of neural language models. First, we draw a formal equivalence between label smoothing, a popular regularization technique for neural language models, and add-𝜆 smoothing. Second, we derive a generalized framework for converting any n-gram smoothing technique into a regularizer compatible with neural language models. Our empirical results find that our novel regularizers are comparable to and, indeed, sometimes outperform label smoothing on language modeling and machine translation.
Current legal outcome prediction models - a staple of legal NLP - do not explain their reasoning. However, to employ these models in the real world, human legal actors need to be able to understand the model’s decisions. In the case of common law, legal practitioners reason towards the outcome of a case by referring to past case law, known as precedent. We contend that precedent is, therefore, a natural way of facilitating explainability for legal NLP models. In this paper, we contribute a novel method for identifying the precedent employed by legal outcome prediction models. Furthermore, by developing a taxonomy of legal precedent, we are able to compare human judges and neural models with respect to the different types of precedent they rely on. We find that while the models learn to predict outcomes reasonably well, their use of precedent is unlike that of human judges.
The introduction of large public legal datasets has brought about a renaissance in legal NLP. Many of these datasets are composed of legal judgments—the product of judges deciding cases. Since ML algorithms learn to model the data they are trained on, several legal NLP models are models of judges. While some have argued for the automation of judges, in this position piece, we argue that automating the role of the judge raises difficult ethical challenges, in particular for common law legal systems. Our argument follows from the social role of the judge in actively shaping the law, rather than merely applying it. Since current NLP models are too far away from having the facilities necessary for this task, they should not be used to automate judges. Furthermore, even in the case that the models could achieve human-level capabilities, there would still be remaining ethical concerns inherent in the automation of the legal process.
Sentiment analysis has become a central tool in various disciplines outside of natural language processing. In particular in applied and domain-specific settings with strong requirements for interpretable methods, dictionary-based approaches are still a popular choice. However, existing dictionaries are often limited in coverage, static once annotation is completed and sentiment scales differ widely; some are discrete others continuous. We propose a Bayesian generative model that learns a composite sentiment dictionary as an interpolation between six existing dictionaries with different scales. We argue that sentiment is a latent concept with intrinsically ranking-based characteristics — the word “excellent” may be ranked more positive than “great” and “okay”, but it is hard to express how much more exactly. This prompts us to enforce an ordinal scale of ordered discrete sentiment values in our dictionary. We achieve this through an ordering transformation in the priors of our model. We evaluate the model intrinsically by imputing missing values in existing dictionaries. Moreover, we conduct extrinsic evaluations through sentiment classification tasks. Finally, we present two extension: first, we present a method to augment dictionary-based approaches with word embeddings to construct sentiment scales along new semantic axes. Second, we demonstrate a Latent Dirichlet Allocation-inspired variant of our model that learns document topics that are ordered by sentiment.
The Bar-Hillel construction is a classic result in formal language theory. It shows, by a simple construction, that the intersection of a context-free language and a regular language is itself context-free. In the construction, the regular language is specified by a finite-state automaton. However, neither the original construction (Bar-Hillel et al., 1961) nor its weighted extension (Nederhof and Satta, 2003) can handle finite-state automata with ε-arcs. While it is possible to remove ε-arcs from a finite-state automaton efficiently without modifying the language, such an operation modifies the automaton’s set of paths. We give a construction that generalizes the Bar- Hillel in the case the desired automaton has ε-arcs, and further prove that our generalized construction leads to a grammar that encodes the structure of both the input automaton and grammar while retaining the asymptotic size of the original construction.
Byte-Pair Encoding (BPE) is a popular algorithm used for tokenizing data in NLP, despite being devised initially as a compression method.BPE appears to be a greedy algorithm at face value, but the underlying optimization problem that BPE seeks to solve has not yet been laid down. We formalize BPE as a combinatorial optimization problem. Via submodular functions, we prove that the iterative greedy version is a 1/sigma*(1-e(-sigma))-approximation of an optimal merge sequence, where sigma is the total backward curvature with respect to the optimal merge sequence. Empirically the lower bound of the approximation is approx0.37.We provide a faster implementation of BPE which improves the runtime complexity from O(NM) to O(N log M), where N is the sequence length and M is the merge count. Finally, we optimize the brute-force algorithm for optimal BPE using memoization.
Sampling-based decoding strategies are widely employed for generating text from probabilistic models, yet standard ancestral sampling often results in text that is degenerate or incoherent. To alleviate this issue, various modifications to a model’s sampling distribution, such as top-p or top-k sampling, have been introduced and are now ubiquitously used in language generation systems. We propose a unified framework for understanding these techniques, which we term sampling adapters. Sampling adapters often lead to qualitatively better text, which raises the question: From a formal perspective, how are they changing the token-level distributions of language generation models? And why do these local changes lead to higher-quality text? We argue that the shift they enforce can be viewed as a trade-off between precision and recall: while the model loses its ability to produce certain strings, its precision rate on desirable text increases. While this trade-off is not reflected in standard metrics of distribution quality (such as perplexity), we find that several precision-emphasizing measures indeed indicate that sampling adapters can lead to probability distributions more aligned with the true distribution. Further, these measures correlate with higher sequence-level quality scores, specifically, Mauve.
We present Earley’s (1970) context-free parsing algorithm as a deduction system, incorporating various known and new speed-ups. In particular, our presentation supports a known worst-case runtime improvement from Earley’s (1970) O(N3|G||R|), which is unworkable for the large grammars that arise in natural language processing, to O(N3|G|), which matches the complexity of CKY on a binarized version of the grammar G. Here N is the length of the sentence, |R| is the number of productions in G, and |G| is the total length of those productions. We also provide a version that achieves runtime of O(N3|M|) with |M| ≤ |G| when the grammar is represented compactly as a single finite-state automaton M (this is partly novel). We carefully treat the generalization to semiring-weighted deduction, preprocessing the grammar like Stolcke (1995) to eliminate the possibility of deduction cycles, and further generalize Stolcke’s method to compute the weights of sentence prefixes. We also provide implementation details for efficient execution, ensuring that on a preprocessed grammar, the semiring-weighted versions of our methods have the same asymptotic runtime and space requirements as the unweighted methods, including sub-cubic runtime on some grammars.
Measuring the intensity of events is crucial for monitoring and tracking armed conflict. Advances in automated event extraction have yielded massive data sets of “who did what to whom” micro-records that enable data-driven approaches to monitoring conflict. The Goldstein scale is a widely-used expert-based measure that scores events on a conflictual–cooperative scale. It is based only on the action category (“what”) and disregards the subject (“who”) and object (“to whom”) of an event, as well as contextual information, like associated casualty count, that should contribute to the perception of an event’s “intensity”. This paper takes a latent variable-based approach to measuring conflict intensity. We introduce a probabilistic generative model that assumes each observed event is associated with a latent intensity class. A novel aspect of this model is that it imposes an ordering on the classes, such that higher-valued classes denote higher levels of intensity. The ordinal nature of the latent variable is induced from naturally ordered aspects of the data (e.g., casualty counts) where higher values naturally indicate higher intensity. We evaluate the proposed model both intrinsically and extrinsically, showing that it obtains comparatively good held-out predictive performance.
Subword tokenization is a key part of most NLP pipelines. However, little is known about why some tokenizer and hyperparameter combinations lead to improved downstream model performance over others. We propose that good tokenizers lead to efficient channel usage, where the channel is the means by which some input is conveyed to the model and efficiency can be quantified in information-theoretic terms as the ratio of the Shannon entropy to the maximum entropy of the subword distribution. Nevertheless, an optimal encoding according to Shannon entropy assigns extremely long codes to low-frequency subwords and very short codes to high-frequency subwords.Defining efficiency in terms of Rényi entropy, on the other hand, penalizes distributions with either very high or very low-frequency subwords.We posit that (1) extremely high-frequency subwords are problematic because their meaning is not distinct and (2) that low-frequency subwords may not appear frequently enough for their meaning to be learned properly; encodings that induce unigram distributions with either can harm model performance. In machine translation, we find that across multiple tokenizers, the Rényi entropy has a very strong correlation with BLEU: 0.82 in comparison to just -0.30 for compressed length.
Weir has defined a hierarchy of language classes whose second member (L2) is generated by tree-adjoining grammars (TAG), linear indexed grammars (LIG), combinatory categorial grammars, and head grammars. The hierarchy is obtained using the mechanism of control, and L2 is obtained using a context-free grammar (CFG) whose derivations are controlled by another CFG. We adapt Weir’s definition of a controllable CFG (called a labeled distinguished CFG) to give a definition of controllable pushdown automata (PDAs), called labeled distinguished PDAs. This yields three new characterizations of L2 as the class of languages generated by PDAs controlling PDAs, PDAs controlling CFGs, and CFGs controlling PDAs. We show that these four formalisms are not only weakly equivalent but equivalent in a stricter sense that we call d-weak equivalence. Furthermore, using an even stricter notion of equivalence called d-strong equivalence, we make precise the intuition that a CFG controlling a CFG is a TAG, a PDA controlling a PDA is an embedded PDA, and a PDA controlling a CFG is a LIG. The fourth member of this family, a CFG controlling a PDA, does not correspond to any kind of automaton we know of, so we invent one and call it a Pushdown Adjoining Automaton (PAA).
Several recent papers claim to have achieved human parity at sentence-level machine translation (MT)—especially between high-resource language pairs. In response, the MT community has, in part, shifted its focus to document-level translation. Translating documents requires a deeper understanding of the structure and meaning of text, which is often captured by various kinds of discourse phenomena such as consistency, coherence, and cohesion. However, this renders conventional sentence-level MT evaluation benchmarks inadequate for evaluating the performance of context-aware MT systems. This paperpresents a new dataset with rich discourse annotations, built upon the large-scale parallel corpus BWB introduced in Jiang et al. (2022a). The new BWB annotation introduces four extra evaluation aspects, i.e., entity, terminology, coreference, and quotation, covering 15,095 entity mentions in both languages. Using these annotations, we systematically investigate the similarities and differences between the discourse structures of source and target languages, and the challenges they pose to MT. We discover that MT outputs differ fundamentally from human translations in terms of their latent discourse structures. This gives us a new perspective on the challenges and opportunities in document-level MT. We make our resource publicly available to spur future research in document-level MT and its generalization to other language translation tasks.
Methods for erasing human-interpretable concepts from neural representations that assume linearity have been found to be tractable and useful. However, the impact of this removal on the behavior of downstream classifiers trained on the modified representations is not fully understood. In this work, we formally define the notion of linear guardedness as the inability of an adversary to predict the concept directly from the representation, and study its implications. We show that, in the binary case, under certain assumptions, a downstream log-linear model cannot recover the erased concept. However, we constructively demonstrate that a multiclass log-linear model can be constructed that indirectly recovers the concept in some cases, pointing to the inherent limitations of linear guardedness as a downstream bias mitigation technique.These findings shed light on the theoretical limitations of linear erasure methods and highlight the need for further research on the connections between intrinsic and extrinsic bias in neural models.
Language modeling, a central task in natural language processing, involves estimating a probability distribution over strings. In most cases, the estimated distribution sums to 1 over all finite strings. However, in some pathological cases, probability mass can “leak” onto the set of infinite sequences. In order to characterize the notion of leakage more precisely, this paper offers a measure-theoretic treatment of language modeling. We prove that many popular language model families are in fact tight, meaning that they will not leak in this sense. We also generalize characterizations of tightness proposed in previous works.
Many popular feature-attribution methods for interpreting deep neural networks rely on computing the gradients of a model’s output with respect to its inputs. While these methods can indicate which input features may be important for the model’s prediction, they reveal little about the inner workings of the model itself. In this paper, we observe that the gradient computation of a model is a special case of a more general formulation using semirings. This observation allows us to generalize the backpropagation algorithm to efficiently compute other interpretable statistics about the gradient graph of a neural network, such as the highest-weighted path and entropy. We implement this generalized algorithm, evaluate it on synthetic datasets to better understand the statistics it computes, and apply it to study BERT’s behavior on the subject–verb number agreement task (SVA). With this method, we (a) validate that the amount of gradient flow through a component of a model reflects its importance to a prediction and (b) for SVA, identify which pathways of the self-attention mechanism are most important.
Multiple algorithms are known for efficiently calculating the prefix probability of a string under a probabilistic context-free grammar (PCFG). Good algorithms for the problem have a runtime cubic in the length of the input string. However, some proposed algorithms are suboptimal with respect to the size of the grammar. This paper proposes a new speed-up of Jelinek and Lafferty’s (1991) algorithm, which runs in O(n3|N|3 + |N|4), where n is the input length and |N| is the number of non-terminals in the grammar. In contrast, our speed-up runs in O(n2|N|3 + n3|N|2).
We introduce a novel dependency parser, the hexatagger, that constructs dependency trees by tagging the words in a sentence with elements from a finite set of possible tags. In contrast to many approaches to dependency parsing, our approach is fully parallelizable at training time, i.e., the structure-building actions needed to build a dependency parse can be predicted in parallel to each other. Additionally, exact decoding is linear in time and space complexity. Furthermore, we derive a probabilistic dependency parser that predicts hexatags using no more than a linear model with features from a pretrained language model, i.e., we forsake a bespoke architecture explicitly designed for the task. Despite the generality and simplicity of our approach, we achieve state-of-the-art performance of 96.4 LAS and 97.4 UAS on the Penn Treebank test set. Additionally, our parser’s linear time complexity and parallelism significantly improve computational efficiency, with a roughly 10-times speed-up over previous state-of-the-art models during decoding.
An increasingly large percentage of natural language processing (NLP) tasks center around the generation of text from probabilistic language models. Despite this trend, techniques for improving or specifying preferences in these generated texts rely mostly on intuition-based heuristics. Further, there lacks a unified presentation of their motivations, practical implementation, successes and pitfalls. Practitioners must, therefore, choose somewhat blindly between generation algorithms—like top-p sampling or beam search—which can lead to wildly different results. At the same time, language generation research continues to criticize and improve the standard toolboxes, further adding entropy to the state of the field. In this tutorial, we will provide a centralized and cohesive discussion of critical considerations when choosing how to generate from a language model. We will cover a wide range of empirically-observed problems (like degradation, hallucination, repetition) and their corresponding proposed algorithmic solutions from recent research (like top-p sampling and its successors). We will then discuss a subset of these algorithms under a unified light; most stochastic generation strategies can be framed as locally adapting the probabilities of a model to avoid failure cases. Finally, we will then cover methods in controlled generation, that go beyond just ensuring coherence to ensure text exhibits specific desired properties. We aim for NLP practitioners and researchers to leave our tutorial with a unified framework which they can use to evaluate and contribute to the latest research in language generation.
Every legal case sets a precedent by developing the law in one of the following two ways. It either expands its scope, in which case it sets positive precedent, or it narrows it, in which case it sets negative precedent. Legal outcome prediction, the prediction of positive outcome, is an increasingly popular task in AI. In contrast, we turn our focus to negative outcomes here, and introduce a new task of negative outcome prediction. We discover an asymmetry in existing models’ ability to predict positive and negative outcomes. Where the state-of-the-art outcome prediction model we used predicts positive outcomes at 75.06 F1, it predicts negative outcomes at only 10.09 F1, worse than a random baseline. To address this performance gap, we develop two new models inspired by the dynamics of a court process. Our first model significantly improves positive outcome prediction score to 77.15 F1 and our second model more than doubles the negative outcome prediction performance to 24.01 F1. Despite this improvement, shifting focus to negative outcomes reveals that there is still much room for improvement for outcome prediction models. https://github.com/valvoda/Negative-Precedent-in-Legal-Outcome-Prediction
Today’s probabilistic language generators fall short when it comes to producing coherent and fluent text despite the fact that the underlying models perform well under standard metrics (e.g., perplexity). This discrepancy has puzzled the language generation community for the last few years. In this work, we posit that the abstraction of natural language generation as a discrete stochastic process—which allows for an information-theoretic analysis—can provide new insights into the behavior of probabilistic language generators, for example, why high-probability texts can be dull or repetitive. Humans use language as a means of communicating information, aiming to do so in a simultaneously efficient and error-minimizing manner; in fact, psycholinguistics research suggests humans choose each word in a string with this subconscious goal in mind. We formally define the set of strings that meet this criterion: Those for which each word has an information content close to the expected information content, namely, the conditional entropy of our model. We then propose a simple and efficient procedure for enforcing this criterion when generating from probabilistic models, which we call locally typical sampling. Automatic and human evaluations show that, in comparison to nucleus and top-k sampling, locally typical sampling offers competitive performance (in both abstractive summarization and story generation) in terms of quality while consistently reducing degenerate repetitions.
Probing has become a go-to methodology for interpreting and analyzing deep neural models in natural language processing. However, there is still a lack of understanding of the limitations and weaknesses of various types of probes. In this work, we suggest a strategy for input-level intervention on naturalistic sentences. Using our approach, we intervene on the morpho-syntactic features of a sentence, while keeping the rest of the sentence unchanged. Such an intervention allows us to causally probe pre-trained models. We apply our naturalistic causal probing framework to analyze the effects of grammatical gender and number on contextualized representations extracted from three pre-trained models in Spanish, the multilingual versions of BERT, RoBERTa, and GPT-2. Our experiments suggest that naturalistic interventions lead to stable estimates of the causal effects of various linguistic properties. Moreover, our experiments demonstrate the importance of naturalistic causal probing when analyzing pre-trained models. https://github.com/rycolab/naturalistic-causal-probing
While natural languages differ widely in both canonical word order and word order flexibility, their word orders still follow shared cross-linguistic statistical patterns, often attributed to functional pressures. In the effort to identify these pressures, prior work has compared real and counterfactual word orders. Yet one functional pressure has been overlooked in such investigations: The uniform information density (UID) hypothesis, which holds that information should be spread evenly throughout an utterance. Here, we ask whether a pressure for UID may have influenced word order patterns cross-linguistically. To this end, we use computational models to test whether real orders lead to greater information uniformity than counterfactual orders. In our empirical study of 10 typologically diverse languages, we find that: (i) among SVO languages, real word orders consistently have greater uniformity than reverse word orders, and (ii) only linguistically implausible counterfactual orders consistently exceed the uniformity of real orders. These findings are compatible with a pressure for information uniformity in the development and usage of natural languages.1
Surprisal theory posits that less-predictable words should take more time to process, with word predictability quantified as surprisal, i.e., negative log probability in context. While evidence supporting the predictions of surprisal theory has been replicated widely, much of it has focused on a very narrow slice of data: native English speakers reading English texts. Indeed, no comprehensive multilingual analysis exists. We address this gap in the current literature by investigating the relationship between surprisal and reading times in eleven different languages, distributed across five language families. Deriving estimates from language models trained on monolingual and multilingual corpora, we test three predictions associated with surprisal theory: (i) whether surprisal is predictive of reading times, (ii) whether expected surprisal, i.e., contextual entropy, is predictive of reading times, and (iii) whether the linking function between surprisal and reading times is linear. We find that all three predictions are borne out crosslinguistically. By focusing on a more diverse set of languages, we argue that these results offer the most robust link to date between information theory and incremental language processing across languages.
Tasks that model the relation between pairs of tokens in a string are a vital part of understanding natural language. Such tasks, in general, require exhaustive pair-wise comparisons of tokens, thus having a quadratic runtime complexity in the length of the string. We show that these exhaustive comparisons can be avoided, and, moreover, the complexity of such tasks can be reduced to linear by casting the relation between tokens as a partial order over the string. Our method predicts real numbers for each token in a string in parallel and sorts the tokens accordingly, resulting in total orders of the tokens in the string. Each total order implies a set of arcs oriented from smaller to greater tokens, sorted by their predicted numbers. The intersection of total orders results in a partial order over the set of tokens in the string, which is then decoded into a directed graph representing the desired linguistic structure. Our experiments on dependency parsing and coreference resolution show that our method achieves state-of-the-art or comparable performance. Moreover, the linear complexity and parallelism of our method double the speed of graph-based coreference resolution models, and bring a 10-times speed-up over graph-based dependency parsers.
Zipf (1935) posited that wordforms are optimized to minimize utterances’ communicative costs. Under the assumption that cost is given by an utterance’s length, he supported this claim by showing that words’ lengths are inversely correlated with their frequencies. Communicative cost, however, can be operationalized in different ways. Piantadosi et al. (2011) claim that cost should be measured as the distance between an utterance’s information rate and channel capacity, which we dub the channel capacity hypothesis (CCH) here. Following this logic, they then proposed that a word’s length should be proportional to the expected value of its surprisal (negative log-probability in context). In this work, we show that Piantadosi et al.’s derivation does not minimize CCH’s cost, but rather a lower bound, which we term CCH-lower. We propose a novel derivation, suggesting an improved way to minimize CCH’s cost. Under this method, we find that a language’s word lengths should instead be proportional to the surprisal’s expectation plus its variance-to-mean ratio. Experimentally, we compare these three communicative cost functions: Zipf’s, CCH-lower , and CCH. Across 13 languages and several experimental settings, we find that length is better predicted by frequency than either of the other hypotheses. In fact, when surprisal’s expectation, or expectation plus variance-to-mean ratio, is estimated using better language models, it leads to worse word length predictions. We take these results as evidence that Zipf’s longstanding hypothesis holds.
The class of tree-adjoining languages can be characterized by various two-level formalisms, consisting of a context-free grammar (CFG) or pushdown automaton (PDA) controlling another CFG or PDA. These four formalisms are equivalent to tree-adjoining grammars (TAG), linear indexed grammars (LIG), pushdown-adjoining automata (PAA), and embedded pushdown automata (EPDA). We define semiring-weighted versions of the above two-level formalisms, and we design new algorithms for computing their stringsums (the weight of all derivations of a string) and allsums (the weight of all derivations). From these, we also immediately obtain stringsum and allsum algorithms for TAG, LIG, PAA, and EPDA. For LIG, our algorithm is more time-efficient by a factor of 𝒪(n|𝒩|) (where n is the string length and |𝒩| is the size of the nonterminal set) and more space-efficient by a factor of 𝒪(|𝛤|) (where 𝛤 is the size of the stack alphabet) than the algorithm of Vijay-Shanker and Weir (1989). For EPDA, our algorithm is both more space-efficient and time-efficient than the algorithm of Alonso et al. (2001) by factors of 𝒪(|𝛤|2) and 𝒪(|𝛤|3), respectively. Finally, we give the first PAA stringsum and allsum algorithms.
This work investigates the computational expressivity of language models (LMs) based on recurrent neural networks (RNNs). Siegelmann and Sontag (1992) famously showed that RNNs with rational weights and hidden states and unbounded computation time are Turing complete. However, LMs define weightings over strings in addition to just (unweighted) language membership and the analysis of the computational power of RNN LMs (RLMs) should reflect this. We extend the Turing completeness result to the probabilistic case, showing how a rationally weighted RLM with unbounded computation time can simulate any deterministic probabilistic Turing machine (PTM) with rationally weighted transitions. Since, in practice, RLMs work in real-time, processing a symbol at every time step, we treat the above result as an upper bound on the expressivity of RLMs. We also provide a lower bound by showing that under the restriction to real-time computation, such models can simulate deterministic real-time rational PTMs.
Surprisal theory (Hale, 2001; Levy, 2008) posits that a word’s reading time is proportional to its surprisal (i.e., to its negative log probability given the proceeding context). Since we are unable to access a word’s ground-truth probability, surprisal theory has been empirically tested using surprisal estimates from language models (LMs). Under the premise that surprisal theory holds, we would expect that higher quality language models provide more powerful predictors of human reading behavior—a conjecture we dub the quality–power (QP) hypothesis. Unfortunately, empirical support for the QP hypothesis is mixed. Some studies in English have found correlations between LM quality and predictive power, but other studies using Japanese data, as well as using larger English LMs, find no such correlations. In this work, we conduct a systematic crosslinguistic assessment of the QP hypothesis. We train LMs from scratch on small- and medium-sized datasets from 13 languages (across five language families) and assess their ability to predict eye tracking data. We find correlations between LM quality and power in eleven of these thirteen languages, suggesting that, within the range of model classes and sizes tested, better language models are indeed better predictors of human language processing behaviors.
Studying language models (LMs) in terms of well-understood formalisms allows us to precisely characterize their abilities and limitations. Previous work has investigated the expressive power of recurrent neural network (RNN) LMs in terms of their capacity to recognize unweighted formal languages. However, LMs do not describe unweighted formal languages—rather, they define probability distributions over strings. In this work, we study what classes of such probability distributions RNN LMs can represent, which allows us to make more direct statements about their capabilities. We show that simple RNNs are equivalent to a subclass of probabilistic finite-state automata, and can thus model a strict subset of probability distributions expressible by finite-state models. Furthermore, we study the space complexity of representing finite-state LMs with RNNs. We show that, to represent an arbitrary deterministic finite-state LM with N states over an alphabet 𝛴, an RNN requires 𝛺\left(N |𝛴|\right) neurons. These results present a first step towards characterizing the classes of distributions RNN LMs can represent and thus help us understand their capabilities and limitations.
Prosody—the suprasegmental component of speech, including pitch, loudness, and tempo—carries critical aspects of meaning. However, the relationship between the information conveyed by prosody vs. by the words themselves remains poorly understood. We use large language models (LLMs) to estimate how much information is redundant between prosody and the words themselves. Using a large spoken corpus of English audiobooks, we extract prosodic features aligned to individual words and test how well they can be predicted from LLM embeddings, compared to non-contextual word embeddings. We find a high degree of redundancy between the information carried by the words and prosodic information across several prosodic features, including intensity, duration, pauses, and pitch contours. Furthermore, a word’s prosodic information is redundant with both the word itself and the context preceding as well as following it. Still, we observe that prosodic features can not be fully predicted from text, suggesting that prosody carries information above and beyond the words. Along with this paper, we release a general-purpose data processing pipeline for quantifying the relationship between linguistic information and extra-linguistic features.
The left-corner transformation (Rosenkrantz and Lewis, 1970) is used to remove left recursion from context-free grammars, which is an important step towards making the grammar parsable top-down with simple techniques. This paper generalizes prior left-corner transformations to support semiring-weighted production rules and to provide finer-grained control over which left corners may be moved. Our generalized left-corner transformation (GLCT) arose from unifying the left-corner transformation and speculation transformation (Eisner and Blatz, 2007), originally for logic programming. Our new transformation and speculation define equivalent weighted languages. Yet, their derivation trees are structurally different in an important way: GLCT replaces left recursion with right recursion, and speculation does not. We also provide several technical results regarding the formal relationships between the outputs of GLCT, speculation, and the original grammar. Lastly, we empirically investigate the efficiency of GLCT for left-recursion elimination from grammars of nine languages. Code: https://github.com/rycolab/left-corner
Pre-trained contextual representations have led to dramatic performance improvements on a range of downstream tasks. Such performance improvements have motivated researchers to quantify and understand the linguistic information encoded in these representations. In general, researchers quantify the amount of linguistic information through probing, an endeavor which consists of training a supervised model to predict a linguistic property directly from the contextual representations. Unfortunately, this definition of probing has been subject to extensive criticism in the literature, and has been observed to lead to paradoxical and counter-intuitive results. In the theoretical portion of this paper, we take the position that the goal of probing ought to be measuring the amount of inductive bias that the representations encode on a specific task. We further describe a Bayesian framework that operationalizes this goal and allows us to quantify the representations’ inductive bias. In the empirical portion of the paper, we apply our framework to a variety of NLP tasks. Our results suggest that our proposed framework alleviates many previous problems found in probing. Moreover, we are able to offer concrete evidence that—for some tasks—fastText can offer a better inductive bias than BERT.
A central quest of probing is to uncover how pre-trained models encode a linguistic property within their representations. An encoding, however, might be spurious—i.e., the model might not rely on it when making predictions. In this paper, we try to find an encoding that the model actually uses, introducing a usage-based probing setup. We first choose a behavioral task which cannot be solved without using the linguistic property. Then, we attempt to remove the property by intervening on the model’s representations. We contend that, if an encoding is used by the model, its removal should harm the performance on the chosen behavioral task. As a case study, we focus on how BERT encodes grammatical number, and on how it uses this encoding to solve the number agreement task. Experimentally, we find that BERT relies on a linear encoding of grammatical number to produce the correct behavioral output. We also find that BERT uses a separate encoding of grammatical number for nouns and verbs. Finally, we identify in which layers information about grammatical number is transferred from a noun to its head verb.
Numerous analyses of reading time (RT) data have been undertaken in the effort to learn more about the internal processes that occur during reading comprehension. However, data measured on words at the end of a sentence–or even clause–is often omitted due to the confounding factors introduced by so-called “wrap-up effects,” which manifests as a skewed distribution of RTs for these words. Consequently, the understanding of the cognitive processes that might be involved in these effects is limited. In this work, we attempt to learn more about these processes by looking for the existence–or absence–of a link between wrap-up effects and information theoretic quantities, such as word and context information content. We find that the information distribution of prior context is often predictive of sentence- and clause-final RTs (while not of sentence-medial RTs), which lends support to several prior hypotheses about the processes involved in wrap-up effects.
When generating natural language from neural probabilistic models, high probability does not always coincide with high quality: It has often been observed that mode-seeking decoding methods, i.e., those that produce high-probability text under the model, lead to unnatural language. On the other hand, the lower-probability text generated by stochastic methods is perceived as more human-like. In this note, we offer an explanation for this phenomenon by analyzing language generation through an information-theoretic lens. Specifically, we posit that human-like language should contain an amount of information (quantified as negative log-probability) that is close to the entropy of the distribution over natural strings. Further, we posit that language with substantially more (or less) information is undesirable. We provide preliminary empirical evidence in favor of this hypothesis; quality ratings of both human and machine-generated text—covering multiple tasks and common decoding strategies—suggest high-quality text has an information content significantly closer to the entropy than we would expect by chance.
Shannon entropy is often a quantity of interest to linguists studying the communicative capacity of human language. However, entropymust typically be estimated from observed data because researchers do not have access to the underlying probability distribution. While entropy estimation is a well-studied problem in other fields, there is not yet a comprehensive exploration of the efficacy of entropy estimators for use with linguistic data. In this work, we fill this void, studying the empirical effectiveness of different entropy estimators for linguistic distributions. In a replication of two recent information-theoretic linguistic studies, we find evidence that the reported effect size is over-estimated due to over-reliance on poor entropy estimators. We end this paper with a concrete recommendation for the entropy estimators that should be used in future linguistic studies.
The Universal Morphology (UniMorph) project is a collaborative effort providing broad-coverage instantiated normalized morphological inflection tables for hundreds of diverse world languages. The project comprises two major thrusts: a language-independent feature schema for rich morphological annotation, and a type-level resource of annotated data in diverse languages realizing that schema. This paper presents the expansions and improvements on several fronts that were made in the last couple of years (since McCarthy et al. (2020)). Collaborative efforts by numerous linguists have added 66 new languages, including 24 endangered languages. We have implemented several improvements to the extraction pipeline to tackle some issues, e.g., missing gender and macrons information. We have amended the schema to use a hierarchical structure that is needed for morphological phenomena like multiple-argument agreement and case stacking, while adding some missing morphological features to make the schema more inclusive. In light of the last UniMorph release, we also augmented the database with morpheme segmentation for 16 languages. Lastly, this new release makes a push towards inclusion of derivational morphology in UniMorph by enriching the data and annotation schema with instances representing derivational processes from MorphyNet.
When generating text from probabilistic models, the chosen decoding strategy has a profound effect on the resulting text. Yet the properties elicited by various decoding strategies do not always transfer across natural language generation tasks. For example, while mode-seeking methods like beam search perform remarkably well for machine translation, they have been observed to lead to incoherent and repetitive text in story generation. Despite such observations, the effectiveness of decoding strategies is often assessed on only a single task. This work—in contrast—provides a comprehensive analysis of the interaction between language generation tasks and decoding strategies. Specifically, we measure changes in attributes of generated text as a function of both decoding strategy and task using human and automatic evaluation. Our results reveal both previously observed and novel findings. For example, the nature of the diversity–quality trade-off in language generation is very task-specific; the length bias often attributed to beam search is not constant across tasks. https://github.com/gianwiher/decoding-NLG
This study describes the structure and the results of the SIGTYP 2022 shared task on the prediction of cognate reflexes from multilingual wordlists. We asked participants to submit systems that would predict words in individual languages with the help of cognate words from related languages. Training and surprise data were based on standardized multilingual wordlists from several language families. Four teams submitted a total of eight systems, including both neural and non-neural systems, as well as systems adjusted to the task and systems using more general settings. While all systems showed a rather promising performance, reflecting the overwhelming regularity of sound change, the best performance throughout was achieved by a system based on convolutional networks originally designed for image restoration.
Recent years have seen a paradigm shift in NLP towards using pretrained language models (PLM) for a wide range of tasks. However, there are many difficult design decisions to represent structures (e.g. tagged text, coreference chains) in a way such that they can be captured by PLMs. Prior work on structured prediction with PLMs typically flattens the structured output into a sequence, which limits the quality of structural information being learned and leads to inferior performance compared to classic discriminative models. In this work, we describe an approach to model structures as sequences of actions in an autoregressive manner with PLMs, allowing in-structure dependencies to be learned without any loss. Our approach achieves the new state-of-the-art on all the structured prediction tasks we looked at, namely, named entity recognition, end-to-end relation extraction, and coreference resolution.
The SIGMORPHON 2022 shared task on morpheme segmentation challenged systems to decompose a word into a sequence of morphemes and covered most types of morphology: compounds, derivations, and inflections. Subtask 1, word-level morpheme segmentation, covered 5 million words in 9 languages (Czech, English, Spanish, Hungarian, French, Italian, Russian, Latin, Mongolian) and received 13 system submissions from 7 teams and the best system averaged 97.29% F1 score across all languages, ranging English (93.84%) to Latin (99.38%). Subtask 2, sentence-level morpheme segmentation, covered 18,735 sentences in 3 languages (Czech, English, Mongolian), received 10 system submissions from 3 teams, and the best systems outperformed all three state-of-the-art subword tokenization methods (BPE, ULM, Morfessor2) by 30.71% absolute. To facilitate error analysis and support any type of future studies, we released all system predictions, the evaluation script, and all gold standard datasets.
The 2022 SIGMORPHON–UniMorph shared task on large scale morphological inflection generation included a wide range of typologically diverse languages: 33 languages from 11 top-level language families: Arabic (Modern Standard), Assamese, Braj, Chukchi, Eastern Armenian, Evenki, Georgian, Gothic, Gujarati, Hebrew, Hungarian, Itelmen, Karelian, Kazakh, Ket, Khalkha Mongolian, Kholosi, Korean, Lamahalot, Low German, Ludic, Magahi, Middle Low German, Old English, Old High German, Old Norse, Polish, Pomak, Slovak, Turkish, Upper Sorbian, Veps, and Xibe. We emphasize generalization along different dimensions this year by evaluating test items with unseen lemmas and unseen features separately under small and large training conditions. Across the five submitted systems and two baselines, the prediction of inflections with unseen features proved challenging, with average performance decreased substantially from last year. This was true even for languages for which the forms were in principle predictable, which suggests that further work is needed in designing systems that capture the various types of generalization required for the world’s languages.
Ethics is one of the longest standing intellectual endeavors of humanity. In recent years, the fields of AI and NLP have attempted to address issues of harmful outcomes in machine learning systems that are made to interface with humans. One recent approach in this vein is the construction of NLP morality models that can take in arbitrary text and output a moral judgment about the situation described. In this work, we offer a critique of such NLP methods for automating ethical decision-making. Through an audit of recent work on computational approaches for predicting morality, we examine the broader issues that arise from such efforts. We conclude with a discussion of how machine ethics could usefully proceed in NLP, by focusing on current and near-future uses of technology, in a way that centers around transparency, democratic values, and allows for straightforward accountability.
Probing is a popular approach to understand what linguistic information is contained in the representations of pre-trained language models. However, the mechanism of selecting the probe model has recently been subject to intense debate, as it is not clear if the probes are merely extracting information or modelling the linguistic property themselves. To address this challenge, this paper introduces a novel model-free approach to probing via prompting, which formulates probing as a prompting task. We conduct experiments on five probing tasks and show that PP is comparable or better at extracting information than diagnostic probes while learning much less on its own. We further combine the probing via prompting approach with pruning to analyze where the model stores the linguistic information in its architecture. Finally, we apply the probing via prompting approach to examine the usefulness of a linguistic property for pre-training by removing the heads that are essential to it and evaluating the resulting model’s performance on language modeling.
Standard automatic metrics, e.g. BLEU, are not reliable for document-level MT evaluation. They can neither distinguish document-level improvements in translation quality from sentence-level ones, nor identify the discourse phenomena that cause context-agnostic translations. This paper introduces a novel automatic metric BlonDe to widen the scope of automatic MT evaluation from sentence to document level. BlonDe takes discourse coherence into consideration by categorizing discourse-related spans and calculating the similarity-based F1 measure of categorized spans. We conduct extensive comparisons on a newly constructed dataset BWB. The experimental results show that BlonDe possesses better selectivity and interpretability at the document-level, and is more sensitive to document-level nuances. In a large-scale human study, BlonDe also achieves significantly higher Pearson’s r correlation with human judgments compared to previous metrics.
The success of multilingual pre-trained models is underpinned by their ability to learn representations shared by multiple languages even in absence of any explicit supervision. However, it remains unclear how these models learn to generalise across languages. In this work, we conjecture that multilingual pre-trained models can derive language-universal abstractions about grammar. In particular, we investigate whether morphosyntactic information is encoded in the same subset of neurons in different languages. We conduct the first large-scale empirical study over 43 languages and 14 morphosyntactic categories with a state-of-the-art neuron-level probe. Our findings show that the cross-lingual overlap between neurons is significant, but its extent may vary across categories and depends on language proximity and pre-training data size.
Many natural language processing tasks, e.g., coreference resolution and semantic role labeling, require selecting text spans and making decisions about them. A typical approach to such tasks is to score all possible spans and greedily select spans for task-specific downstream processing. This approach, however, does not incorporate any inductive bias about what sort of spans ought to be selected, e.g., that selected spans tend to be syntactic constituents. In this paper, we propose a novel grammar-based structured span selection model which learns to make use of the partial span-level annotation provided for such problems. Compared to previous approaches, our approach gets rid of the heuristic greedy span selection scheme, allowing us to model the downstream task on an optimal set of spans. We evaluate our model on two popular span prediction tasks: coreference resolution and semantic role labeling; and show improvements on both.
Significance testing—especially the paired-permutation test—has played a vital role in developing NLP systems to provide confidence that the difference in performance between two systems (i.e., the test statistic) is not due to luck. However, practitioners rely on Monte Carlo approximation to perform this test due to a lack of a suitable exact algorithm. In this paper, we provide an efficient exact algorithm for the paired-permutation test for a family of structured test statistics. Our algorithm runs in 𝒪(G N (log GN )(log N)) time where N is the dataset size and G is the range of the test statistic. We found that our exact algorithm was 10x faster than the Monte Carlo approximation with 20000 samples on a common dataset
Despite significant progress in the quality of language generated from abstractive summarization models, these models still exhibit the tendency to hallucinate, i.e., output content not supported by the source document. A number of works have tried to fix—or at least uncover the source of—the problem with limited success. In this paper, we identify a simple criterion under which models are significantly more likely to assign more probability to hallucinated content during generation: high model uncertainty. This finding offers a potential explanation for hallucinations: models default to favoring text with high marginal probability, i.e., high-frequency occurrences in the training set, when uncertain about a continuation. It also motivates possible routes for real-time intervention during decoding to prevent such hallucinations. We propose a decoding strategy that switches to optimizing for pointwise mutual information of the source and target token—rather than purely the probability of the target token—when the model exhibits uncertainty. Experiments on the dataset show that our method decreases the probability of hallucinated tokens while maintaining the Rouge and BERT-S scores of top-performing decoding strategies.
The representation space of neural models for textual data emerges in an unsupervised manner during training. Understanding how human-interpretable concepts, such as gender, are encoded in these representations would improve the ability of users to control the content of these representations and analyze the working of the models that rely on them. One prominent approach to the control problem is the identification and removal of linear concept subspaces – subspaces in the representation space that correspond to a given concept. While those are tractable and interpretable, neural network do not necessarily represent concepts in linear subspaces. We propose a kernelization of the recently-proposed linear concept-removal objective, and show that it is effective in guarding against the ability of certain nonlinear adversaries to recover the concept. Interestingly, our findings suggest that the division between linear and nonlinear models is overly simplistic: when considering the concept of binary gender and its neutralization, we do not find a single kernel space that exclusively contains all the concept-related information. It is therefore challenging to protect against all nonlinear adversaries at once.
Weighted finite-state automata (WSFAs) arecommonly used in NLP. Failure transitions area useful extension for compactly representingbackoffs or interpolation in n-gram modelsand CRFs, which are special cases of WFSAs.Unfortunately, applying standard algorithmsfor computing the pathsum requires expand-ing these compact failure transitions. As aresult, na ̈ıve computation of the pathsum inacyclic WFSAs with failure transitions runs inO(|Q|2|Σ|) (O(|Q||Σ|) for deterministic WF-SAs) while the equivalent algorithm in normalWFSAs runs in O(|E|), where E representsthe set of transitions, Q the set of states, andΣ the alphabet. In this work, we present moreefficient algorithms for computing the pathsumin sparse acyclic WFSAs, i.e., WFSAs with av-erage out symbol fraction s ≪ 1. In those,backward runs in O(s|Q||Σ|). We proposean algorithm for semiring-weighted automatawhich runs in O(|E| + s|Σ||Q||Tmax| log |Σ|),where |Tmax| is the size of the largest con-nected component of failure transitions. Ad-ditionally, we propose faster algorithms fortwo specific cases. For ring-weighted WF-SAs we propose an algorithm with complex-ity O(|E| + s|Σ||Q||πmax|), where |πmax| de-notes the longest path length of failure transi-tions stemming from q and Σ(q) the set of sym-bols on the outgoing transitions from q. Forsemiring-weighted WFSAs whose failure tran-sition topology satisfies a condition exemplifiedby CRFs, we propose an algorithm with com-plexity O(|E| + s|Σ||Q| log |Σ|).
There are many proposals to reduce constituency parsing to tagging. To figure out what these approaches have in common, we offer a unifying pipeline, which consists of three steps: linearization, learning, and decoding. We prove that classic shift–reduce parsing can be reduced to tetratagging—the state-of-the-art constituency tagger—under two assumptions: right-corner transformation in the linearization step and factored scoring in the learning step. We ask what is the most critical factor that makes parsing-as-tagging methods accurate while being efficient. To answer this question, we empirically evaluate a taxonomy of tagging pipelines with different choices of linearizers, learners, and decoders. Based on the results in English as well as a set of 8 typologically diverse languages, we conclude that the linearization of the derivation tree and its alignment with the input sequence is the most critical factor in achieving accurate parsers as taggers.
Weighted pushdown automata (WPDAs) are at the core of many natural language processing tasks, like syntax-based statistical machine translation and transition-based dependency parsing. As most existing dynamic programming algorithms are designed for context-free grammars (CFGs), algorithms for PDAs often resort to a PDA-to-CFG conversion. In this paper, we develop novel algorithms that operate directly on WPDAs. Our algorithms are inspired by Lang’s algorithm, but use a more general definition of pushdown automaton and either reduce the space requirements by a factor of |Gamma| (the size of the stack alphabet) or reduce the runtime by a factor of more than |Q| (the number of states). When run on the same class of PDAs as Lang’s algorithm, our algorithm is both more space-efficient by a factor of |Gamma| and more time-efficient by a factor of |Q| x |Gamma|.
The ability to generalize compositionally is key to understanding the potentially infinite number of sentences that can be constructed in a human language from only a finite number of words. Investigating whether NLP models possess this ability has been a topic of interest: SCAN (Lake and Baroni, 2018) is one task specifically proposed to test for this property. Previous work has achieved impressive empirical results using a group-equivariant neural network that naturally encodes a useful inductive bias for SCAN (Gordon et al., 2020). Inspired by this, we introduce a novel group-equivariant architecture that incorporates a group-invariant hard alignment mechanism. We find that our network’s structure allows it to develop stronger equivariance properties than existing group-equivariant approaches. We additionally find that it outperforms previous group-equivariant networks empirically on the SCAN task. Our results suggest that integrating group-equivariance into a variety of neural architectures is a potentially fruitful avenue of research, and demonstrate the value of careful analysis of the theoretical properties of such architectures.
Recombining known primitive concepts into larger novel combinations is a quintessentially human cognitive capability. Whether large neural models in NLP acquire this ability while learning from data is an open question. In this paper, we look at this problem from the perspective of formal languages. We use deterministic finite-state transducers to make an unbounded number of datasets with controllable properties governing compositionality. By randomly sampling over many transducers, we explore which of their properties (number of states, alphabet size, number of transitions etc.) contribute to learnability of a compositional relation by a neural network. In general, we find that the models either learn the relations completely or not at all. The key is transition coverage, setting a soft learnability limit at 400 examples per transition.
Computational models of human language often involve combinatorial problems. For instance, a probabilistic parser may marginalize over exponentially many trees to make predictions. Algorithms for such problems often employ dynamic programming and are not always unique. Finding one with optimal asymptotic runtime can be unintuitive, time-consuming, and error-prone. Our work aims to automate this laborious process. Given an initial correct declarative program, we search for a sequence of semantics-preserving transformations to improve its running time as much as possible. To this end, we describe a set of program transformations, a simple metric for assessing the efficiency of a transformed program, and a heuristic search procedure to improve this metric. We show that in practice, automated search—like the mental search performed by human programmers—can find substantial improvements to the initial program. Empirically, we show that many speed-ups described in the NLP literature could have been discovered automatically by our system.
Large pre-trained language models have repeatedly shown their ability to produce fluent text. Yet even when starting from a prompt, generation can continue in many plausible directions. Current decoding methods with the goal of controlling generation, e.g., to ensure specific words are included, either require additional models or fine-tuning, or work poorly when the task at hand is semantically unconstrained, e.g., story generation. In this work, we present a plug-and-play decoding method for controlled language generation that is so simple and intuitive, it can be described in a single sentence: given a topic or keyword, we add a shift to the probability distribution over our vocabulary towards semantically similar words. We show how annealing this distribution can be used to impose hard constraints on language generation, something no other plug-and-play method is currently able to do with SOTA language generators. Despite the simplicity of this approach, we see it works incredibly well in practice: decoding from GPT-2 leads to diverse and fluent sentences while guaranteeing the appearance of given guide words. We perform two user studies, revealing that (1) our method outperforms competing methods in human evaluations; and (2) forcing the guide words to appear in the generated text has no impact on the fluency of the generated text.
Analysing whether neural language models encode linguistic information has become popular in NLP. One method of doing so, which is frequently cited to support the claim that models like BERT encode syntax, is called probing; probes are small supervised models trained to extract linguistic information from another model’s output. If a probe is able to predict a particular structure, it is argued that the model whose output it is trained on must have implicitly learnt to encode it. However, drawing a generalisation about a model’s linguistic knowledge about a specific phenomena based on what a probe is able to learn may be problematic: in this work, we show that semantic cues in training data means that syntactic probes do not properly isolate syntax. We generate a new corpus of semantically nonsensical but syntactically well-formed Jabberwocky sentences, which we use to evaluate two probes trained on normal data. We train the probes on several popular language models (BERT, GPT-2, and RoBERTa), and find that in all settings they perform worse when evaluated on these data, for one probe by an average of 15.4 UUAS points absolute. Although in most cases they still outperform the baselines, their lead is reduced substantially, e.g. by 53% in the case of BERT for one probe. This begs the question: what empirical scores constitute knowing syntax?
Probes are models devised to investigate the encoding of knowledge—e.g. syntactic structure—in contextual representations. Probes are often designed for simplicity, which has led to restrictions on probe design that may not allow for the full exploitation of the structure of encoded information; one such restriction is linearity. We examine the case of a structural probe (Hewitt and Manning, 2019), which aims to investigate the encoding of syntactic structure in contextual representations through learning only linear transformations. By observing that the structural probe learns a metric, we are able to kernelize it and develop a novel non-linear variant with an identical number of parameters. We test on 6 languages and find that the radial-basis function (RBF) kernel, in conjunction with regularization, achieves a statistically significant improvement over the baseline in all languages—implying that at least part of the syntactic knowledge is encoded non-linearly. We conclude by discussing how the RBF kernel resembles BERT’s self-attention layers and speculate that this resemblance leads to the RBF-based probe’s stronger performance.
In common law, the outcome of a new case is determined mostly by precedent cases, rather than by existing statutes. However, how exactly does the precedent influence the outcome of a new case? Answering this question is crucial for guaranteeing fair and consistent judicial decision-making. We are the first to approach this question computationally by comparing two longstanding jurisprudential views; Halsbury’s, who believes that the arguments of the precedent are the main determinant of the outcome, and Goodhart’s, who believes that what matters most is the precedent’s facts. We base our study on the corpus of legal cases from the European Court of Human Rights (ECtHR), which allows us to access not only the case itself, but also cases cited in the judges’ arguments (i.e. the precedent cases). Taking an information-theoretic view, and modelling the question as a case out-come classification task, we find that the precedent’s arguments share 0.38 nats of information with the case’s outcome, whereas precedent’s facts only share 0.18 nats of information (i.e.,58% less); suggesting Halsbury’s view may be more accurate in this specific court. We found however in a qualitative analysis that there are specific statues where Goodhart’s view dominates, and present some evidence these are the ones where the legal concept at hand is less straightforward.
This work presents an information-theoretic operationalisation of cross-linguistic non-arbitrariness. It is not a new idea that there are small, cross-linguistic associations between the forms and meanings of words. For instance, it has been claimed (Blasi et al., 2016) that the word for “tongue” is more likely than chance to contain the phone [l]. By controlling for the influence of language family and geographic proximity within a very large concept-aligned, cross-lingual lexicon, we extend methods previously used to detect within language non-arbitrariness (Pimentel et al., 2019) to measure cross-linguistic associations. We find that there is a significant effect of non-arbitrariness, but it is unsurprisingly small (less than 0.5% on average according to our information-theoretic estimate). We also provide a concept-level analysis which shows that a quarter of the concepts considered in our work exhibit a significant level of cross-linguistic non-arbitrariness. In sum, the paper provides new methods to detect cross-linguistic associations at scale, and confirms their effects are minor.
The mapping of lexical meanings to wordforms is a major feature of natural languages. While usage pressures might assign short words to frequent meanings (Zipf’s law of abbreviation), the need for a productive and open-ended vocabulary, local constraints on sequences of symbols, and various other factors all shape the lexicons of the world’s languages. Despite their importance in shaping lexical structure, the relative contributions of these factors have not been fully quantified. Taking a coding-theoretic view of the lexicon and making use of a novel generative statistical model, we define upper bounds for the compressibility of the lexicon under various constraints. Examining corpora from 7 typologically diverse languages, we use those upper bounds to quantify the lexicon’s optimality and to explore the relative costs of major constraints on natural codes. We find that (compositional) morphology and graphotactics can sufficiently account for most of the complexity of natural codes—as measured by code length.
Beam search is the default decoding strategy for many sequence generation tasks in NLP. The set of approximate K-best items returned by the algorithm is a useful summary of the distribution for many applications; however, the candidates typically exhibit high overlap and may give a highly biased estimate for expectations under our model. These problems can be addressed by instead using stochastic decoding strategies. In this work, we propose a new method for turning beam search into a stochastic process: Conditional Poisson stochastic beam search. Rather than taking the maximizing set at each iteration, we sample K candidates without replacement according to the conditional Poisson sampling design. We view this as a more natural alternative to Kool et al. (2019)’s stochastic beam search (SBS). Furthermore, we show how samples generated under the CPSBS design can be used to build consistent estimators and sample diverse sets from sequence models. In our experiments, we observe CPSBS produces lower variance and more efficient estimators than SBS, even showing improvements in high entropy settings.
While there exist scores of natural languages, each with its unique features and idiosyncrasies, they all share a unifying theme: enabling human communication. We may thus reasonably predict that human cognition shapes how these languages evolve and are used. Assuming that the capacity to process information is roughly constant across human populations, we expect a surprisal–duration trade-off to arise both across and within languages. We analyse this trade-off using a corpus of 600 languages and, after controlling for several potential confounds, we find strong supporting evidence in both settings. Specifically, we find that, on average, phones are produced faster in languages where they are less surprising, and vice versa. Further, we confirm that more surprising phones are longer, on average, in 319 languages out of the 600. We thus conclude that there is strong evidence of a surprisal–duration trade-off in operation, both across and within the world’s languages.
The uniform information density (UID) hypothesis posits a preference among language users for utterances structured such that information is distributed uniformly across a signal. While its implications on language production have been well explored, the hypothesis potentially makes predictions about language comprehension and linguistic acceptability as well. Further, it is unclear how uniformity in a linguistic signal—or lack thereof—should be measured, and over which linguistic unit, e.g., the sentence or language level, this uniformity should hold. Here we investigate these facets of the UID hypothesis using reading time and acceptability data. While our reading time results are generally consistent with previous work, they are also consistent with a weakly super-linear effect of surprisal, which would be compatible with UID’s predictions. For acceptability judgments, we find clearer evidence that non-uniformity in information density is predictive of lower acceptability. We then explore multiple operationalizations of UID, motivated by different interpretations of the original hypothesis, and analyze the scope over which the pressure towards uniformity is exerted. The explanatory power of a subset of the proposed operationalizations suggests that the strongest trend may be a regression towards a mean surprisal across the language, rather than the phrase, sentence, or document—a finding that supports a typical interpretation of UID, namely that it is the byproduct of language users maximizing the use of a (hypothetical) communication channel.
Pimentel et al. (2020) recently analysed probing from an information-theoretic perspective. They argue that probing should be seen as approximating a mutual information. This led to the rather unintuitive conclusion that representations encode exactly the same information about a target task as the original sentences. The mutual information, however, assumes the true probability distribution of a pair of random variables is known, leading to unintuitive results in settings where it is not. This paper proposes a new framework to measure what we term Bayesian mutual information, which analyses information from the perspective of Bayesian agents—allowing for more intuitive findings in scenarios with finite data. For instance, under Bayesian MI we have that data can add information, processing can help, and information can hurt, which makes it more intuitive for machine learning applications. Finally, we apply our framework to probing where we believe Bayesian mutual information naturally operationalises ease of extraction by explicitly limiting the available background knowledge to solve a task.
Understanding the origins of militarized conflict is a complex, yet important undertaking. Existing research seeks to build this understanding by considering bi-lateral relationships between entity pairs (dyadic causes) and multi-lateral relationships among multiple entities (systemic causes). The aim of this work is to compare these two causes in terms of how they correlate with conflict between two entities. We do this by devising a set of textual and graph-based features which represent each of the causes. The features are extracted from Wikipedia and modeled as a large graph. Nodes in this graph represent entities connected by labeled edges representing ally or enemy-relationships. This allows casting the problem as an edge classification task, which we term dyad classification. We propose and evaluate classifiers to determine if a particular pair of entities are allies or enemies. Our results suggest that our systemic features might be slightly better correlates of conflict. Further, we find that Wikipedia articles of allies are semantically more similar than enemies.
Homophony’s widespread presence in natural languages is a controversial topic. Recent theories of language optimality have tried to justify its prevalence, despite its negative effects on cognitive processing time, e.g., Piantadosi et al. (2012) argued homophony enables the reuse of efficient wordforms and is thus beneficial for languages. This hypothesis has recently been challenged by Trott and Bergen (2020), who posit that good wordforms are more often homophonous simply because they are more phonotactically probable. In this paper, we join in on the debate. We first propose a new information-theoretic quantification of a language’s homophony: the sample Rényi entropy. Then, we use this quantification to revisit Trott and Bergen’s claims. While their point is theoretically sound, a specific methodological issue in their experiments raises doubts about their results. After addressing this issue, we find no clear pressure either towards or against homophony—a much more nuanced result than either Piantadosi et al.’s or Trott and Bergen’s findings.
Probabilistic distributions over spanning trees in directed graphs are a fundamental model of dependency structure in natural language processing, syntactic dependency trees. In NLP, dependency trees often have an additional root constraint: only one edge may emanate from the root. However, no sampling algorithm has been presented in the literature to account for this additional constraint. In this paper, we adapt two spanning tree sampling algorithms to faithfully sample dependency trees from a graph subject to the root constraint. Wilson (1996(’s sampling algorithm has a running time of O(H) where H is the mean hitting time of the graph. Colbourn (1996)’s sampling algorithm has a running time of O(Nˆ3), which is often greater than the mean hitting time of a directed graph. Additionally, we build upon Colbourn’s algorithm and present a novel extension that can sample K trees without replacement in O(K Nˆ3 + Kˆ2 N) time. To the best of our knowledge, no algorithm has been given for sampling spanning trees without replacement from a directed graph.
Psycholinguistic studies of human word processing and lexical access provide ample evidence of the preferred nature of word-initial versus word-final segments, e.g., in terms of attention paid by listeners (greater) or the likelihood of reduction by speakers (lower). This has led to the conjecture—as in Wedel et al. (2019b), but common elsewhere—that languages have evolved to provide more information earlier in words than later. Information-theoretic methods to establish such tendencies in lexicons have suffered from several methodological shortcomings that leave open the question of whether this high word-initial informativeness is actually a property of the lexicon or simply an artefact of the incremental nature of recognition. In this paper, we point out the confounds in existing methods for comparing the informativeness of segments early in the word versus later in the word, and present several new measures that avoid these confounds. When controlling for these confounds, we still find evidence across hundreds of languages that indeed there is a cross-linguistic tendency to front-load information in words.
Neural sequence-to-sequence models are currently the predominant choice for language generation tasks. Yet, on word-level tasks, exact inference of these models reveals the empty string is often the global optimum. Prior works have speculated this phenomenon is a result of the inadequacy of neural models for language generation. However, in the case of morphological inflection, we find that the empty string is almost never the most probable solution under the model. Further, greedy search often finds the global optimum. These observations suggest that the poor calibration of many neural models may stem from characteristics of a specific subset of tasks rather than general ill-suitedness of such models for language generation.
The transformer has been shown to outperform recurrent neural network-based sequence-to-sequence models in various word-level NLP tasks. Yet for character-level transduction tasks, e.g. morphological inflection generation and historical text normalization, there are few works that outperform recurrent models using the transformer. In an empirical study, we uncover that, in contrast to recurrent sequence-to-sequence models, the batch size plays a crucial role in the performance of the transformer on character-level tasks, and we show that with a large enough batch size, the transformer does indeed outperform recurrent models. We also introduce a simple technique to handle feature-guided character-level transduction that further improves performance. With these insights, we achieve state-of-the-art performance on morphological inflection and historical text normalization. We also show that the transformer outperforms a strong baseline on two other character-level transduction tasks: grapheme-to-phoneme conversion and transliteration.
This year’s iteration of the SIGMORPHON Shared Task on morphological reinflection focuses on typological diversity and cross-lingual variation of morphosyntactic features. In terms of the task, we enrich UniMorph with new data for 32 languages from 13 language families, with most of them being under-resourced: Kunwinjku, Classical Syriac, Arabic (Modern Standard, Egyptian, Gulf), Hebrew, Amharic, Aymara, Magahi, Braj, Kurdish (Central, Northern, Southern), Polish, Karelian, Livvi, Ludic, Veps, Võro, Evenki, Xibe, Tuvan, Sakha, Turkish, Indonesian, Kodi, Seneca, Asháninka, Yanesha, Chukchi, Itelmen, Eibela. We evaluate six systems on the new data and conduct an extensive error analysis of the systems’ predictions. Transformer-based models generally demonstrate superior performance on the majority of languages, achieving >90% accuracy on 65% of them. The languages on which systems yielded low accuracy are mainly under-resourced, with a limited amount of data. Most errors made by the systems are due to allomorphy, honorificity, and form variation. In addition, we observe that systems especially struggle to inflect multiword lemmas. The systems also produce misspelled forms or end up in repetitive loops (e.g., RNN-based models). Finally, we report a large drop in systems’ performance on previously unseen lemmas.
While language identification is a fundamental speech and language processing task, for many languages and language families it remains a challenging task. For many low-resource and endangered languages this is in part due to resource availability: where larger datasets exist, they may be single-speaker or have different domains than desired application scenarios, demanding a need for domain and speaker-invariant language identification systems. This year’s shared task on robust spoken language identification sought to investigate just this scenario: systems were to be trained on largely single-speaker speech from one domain, but evaluated on data in other domains recorded from speakers under different recording circumstances, mimicking realistic low-resource scenarios. We see that domain and speaker mismatch proves very challenging for current methods which can perform above 95% accuracy in-domain, which domain adaptation can address to some degree, but that these conditions merit further investigation to make spoken language identification accessible in many scenarios.
We use large-scale corpora in six different gendered languages, along with tools from NLP and information theory, to test whether there is a relationship between the grammatical genders of inanimate nouns and the adjectives used to describe those nouns. For all six languages, we find that there is a statistically significant relationship. We also find that there are statistically significant relationships between the grammatical genders of inanimate nouns and the verbs that take those nouns as direct objects, as indirect objects, and as subjects. We defer deeper investigation of these relationships for future work.
Most combinations of NLP tasks and language varieties lack in-domain examples for supervised training because of the paucity of annotated data. How can neural models make sample-efficient generalizations from task–language combinations with available data to low-resource ones? In this work, we propose a Bayesian generative model for the space of neural parameters. We assume that this space can be factorized into latent variables for each language and each task. We infer the posteriors over such latent variables based on data from seen task–language combinations through variational inference. This enables zero-shot classification on unseen combinations at prediction time. For instance, given training data for named entity recognition (NER) in Vietnamese and for part-of-speech (POS) tagging in Wolof, our model can perform accurate predictions for NER in Wolof. In particular, we experiment with a typologically diverse sample of 33 languages from 4 continents and 11 families, and show that our model yields comparable or better results than state-of-the-art, zero-shot cross-lingual transfer methods. Our code is available at github.com/cambridgeltl/parameter-factorization.
We give a general framework for inference in spanning tree models. We propose unified algorithms for the important cases of first-order expectations and second-order expectations in edge-factored, non-projective spanning-tree models. Our algorithms exploit a fundamental connection between gradients and expectations, which allows us to derive efficient algorithms. These algorithms are easy to implement with or without automatic differentiation software. We motivate the development of our framework with several cautionary tales of previous research, which has developed numerous inefficient algorithms for computing expectations and their gradients. We demonstrate how our framework efficiently computes several quantities with known algorithms, including the expected attachment score, entropy, and generalized expectation criteria. As a bonus, we give algorithms for quantities that are missing in the literature, including the KL divergence. In all cases, our approach matches the efficiency of existing algorithms and, in several cases, reduces the runtime complexity by a factor of the sentence length. We validate the implementation of our framework through runtime experiments. We find our algorithms are up to 15 and 9 times faster than previous algorithms for computing the Shannon entropy and the gradient of the generalized expectation objective, respectively.
Large-scale pretraining and task-specific fine- tuning is now the standard methodology for many tasks in computer vision and natural language processing. Recently, a multitude of methods have been proposed for pretraining vision and language BERTs to tackle challenges at the intersection of these two key areas of AI. These models can be categorized into either single-stream or dual-stream encoders. We study the differences between these two categories, and show how they can be unified under a single theoretical framework. We then conduct controlled experiments to discern the empirical differences between five vision and language BERTs. Our experiments show that training data and hyperparameters are responsible for most of the differences between the reported results, but they also reveal that the embedding layer plays a crucial role in these massive models.
Multi-head attention, a collection of several attention mechanisms that independently attend to different parts of the input, is the key ingredient in the Transformer. Recent work has shown, however, that a large proportion of the heads in a Transformer’s multi-head attention mechanism can be safely pruned away without significantly harming the performance of the model; such pruning leads to models that are noticeably smaller and faster in practice. Our work introduces a new head pruning technique that we term differentiable subset pruning. ntuitively, our method learns per- head importance variables and then enforces a user-specified hard constraint on the number of unpruned heads. he importance variables are learned via stochastic gradient descent. e conduct experiments on natural language inference and machine translation; we show that differentiable subset pruning performs comparably or better than previous works while offering precise control of the sparsity level.1
Since language models are used to model a wide variety of languages, it is natural to ask whether the neural architectures used for the task have inductive biases towards modeling particular types of languages. Investigation of these biases has proved complicated due to the many variables that appear in the experimental setup. Languages vary in many typological dimensions, and it is difficult to single out one or two to investigate without the others acting as confounders. We propose a novel method for investigating the inductive biases of language models using artificial languages. These languages are constructed to allow us to create parallel corpora across languages that differ only in the typological feature being investigated, such as word order. We then use them to train and test language models. This constitutes a fully controlled causal framework, and demonstrates how grammar engineering can serve as a useful tool for analyzing neural models. Using this method, we find that commonly used neural architectures exhibit different inductive biases: LSTMs display little preference with respect to word ordering, while transformers display a clear preference for some orderings over others. Further, we find that neither the inductive bias of the LSTM nor that of the transformer appear to reflect any tendencies that we see in attested natural languages.
The connection between the maximum spanning tree in a directed graph and the best dependency tree of a sentence has been exploited by the NLP community. However, for many dependency parsing schemes, an important detail of this approach is that the spanning tree must have exactly one edge emanating from the root. While work has been done to efficiently solve this problem for finding the one-best dependency tree, no research has attempted to extend this solution to finding the K-best dependency trees. This is arguably a more important extension as a larger proportion of decoded trees will not be subject to the root constraint of dependency trees. Indeed, we show that the rate of root constraint violations increases by an average of 13 times when decoding with K=50 as opposed to K=1. In this paper, we provide a simplification of the K-best spanning tree algorithm of Camerini et al. (1980). Our simplification allows us to obtain a constant time speed-up over the original algorithm. Furthermore, we present a novel extension of the algorithm for decoding the K-best dependency trees of a graph which are subject to a root constraint.
The uniform information density (UID) hypothesis, which posits that speakers behaving optimally tend to distribute information uniformly across a linguistic signal, has gained traction in psycholinguistics as an explanation for certain syntactic, morphological, and prosodic choices. In this work, we explore whether the UID hypothesis can be operationalized as an inductive bias for statistical language modeling. Specifically, we augment the canonical MLE objective for training language models with a regularizer that encodes UID. In experiments on ten languages spanning five language families, we find that using UID regularization consistently improves perplexity in language models, having a larger effect when training data is limited. Moreover, via an analysis of generated sequences, we find that UID-regularized language models have other desirable properties, e.g., they generate text that is more lexically diverse. Our results not only suggest that UID is a reasonable inductive bias for language modeling, but also provide an alternative validation of the UID hypothesis using modern-day NLP tools.
We propose an alternate approach to quantifying how well language models learn natural language: we ask how well they match the statistical tendencies of natural language. To answer this question, we analyze whether text generated from language models exhibits the statistical tendencies present in the human-generated text on which they were trained. We provide a framework–paired with significance tests–for evaluating the fit of language models to these trends. We find that neural language models appear to learn only a subset of the tendencies considered, but align much more closely with empirical trends than proposed theoretical distributions (when present). Further, the fit to different distributions is highly-dependent on both model architecture and generation strategy. As concrete examples, text generated under the nucleus sampling scheme adheres more closely to the type–token relationship of natural language than text produced using standard ancestral sampling; text from LSTMs reflects the natural language distributions over length, stopwords, and symbols surprisingly well.
Beam search is a go-to strategy for decoding neural sequence models. The algorithm can naturally be viewed as a subset optimization problem, albeit one where the corresponding set function does not reflect interactions between candidates. Empirically, this leads to sets often exhibiting high overlap, e.g., strings may differ by only a single word. Yet in use-cases that call for multiple solutions, a diverse or representative set is often desired. To address this issue, we propose a reformulation of beam search, which we call determinantal beam search. Determinantal beam search has a natural relationship to determinantal point processes (DPPs), models over sets that inherently encode intra-set interactions. By posing iterations in beam search as a series of subdeterminant maximization problems, we can turn the algorithm into a diverse subset selection process. In a case study, we use the string subsequence kernel to explicitly encourage n-gram coverage in text generated from a sequence model. We observe that our algorithm offers competitive performance against other diverse set generation strategies in the context of language generation, while providing a more general approach to optimizing for diversity.
Sparse attention has been claimed to increase model interpretability under the assumption that it highlights influential inputs. Yet the attention distribution is typically over representations internal to the model rather than the inputs themselves, suggesting this assumption may not have merit. We build on the recent work exploring the interpretability of attention; we design a set of experiments to help us understand how sparsity affects our ability to use attention as an explainability tool. On three text classification tasks, we verify that only a weak relationship between inputs and co-indexed intermediate representations exists—under sparse attention and otherwise. Further, we do not find any plausible mappings from sparse attention distributions to a sparse set of influential inputs through other avenues. Rather, we observe in this setting that inducing sparsity may make it less plausible that attention can be used as a tool for understanding model behavior.
Weighted finite-state machines are a fundamental building block of NLP systems. They have withstood the test of time—from their early use in noisy channel models in the 1990s up to modern-day neurally parameterized conditional random fields. This work examines the computation of higher-order derivatives with respect to the normalization constant for weighted finite-state machines. We provide a general algorithm for evaluating derivatives of all orders, which has not been previously described in the literature. In the case of second-order derivatives, our scheme runs in the optimal O(Aˆ2 Nˆ4) time where A is the alphabet size and N is the number of states. Our algorithm is significantly faster than prior algorithms. Additionally, our approach leads to a significantly faster algorithm for computing second-order expectations, such as covariance matrices and gradients of first-order expectations.
Typological knowledge bases (KBs) such as WALS (Dryer and Haspelmath, 2013) contain information about linguistic properties of the world’s languages. They have been shown to be useful for downstream applications, including cross-lingual transfer learning and linguistic probing. A major drawback hampering broader adoption of typological KBs is that they are sparsely populated, in the sense that most languages only have annotations for some features, and skewed, in that few features have wide coverage. As typological features often correlate with one another, it is possible to predict them and thus automatically populate typological KBs, which is also the focus of this shared task. Overall, the task attracted 8 submissions from 5 teams, out of which the most successful methods make use of such feature correlations. However, our error analysis reveals that even the strongest submitted systems struggle with predicting feature values for languages where few features are known.
A broad goal in natural language processing (NLP) is to develop a system that has the capacity to process any natural language. Most systems, however, are developed using data from just one language such as English. The SIGMORPHON 2020 shared task on morphological reinflection aims to investigate systems’ ability to generalize across typologically distinct languages, many of which are low resource. Systems were developed using data from 45 languages and just 5 language families, fine-tuned with data from an additional 45 languages and 10 language families (13 in total), and evaluated on all 90 languages. A total of 22 systems (19 neural) from 10 teams were submitted to the task. All four winning systems were neural (two monolingual transformers and two massively multilingual RNN-based models with gated attention). Most teams demonstrate utility of data hallucination and augmentation, ensembles, and multilingual training for low-resource languages. Non-neural learners and manually designed grammars showed competitive and even superior performance on some languages (such as Ingrian, Tajik, Tagalog, Zarma, Lingala), especially with very limited data. Some language families (Afro-Asiatic, Niger-Congo, Turkic) were relatively easy for most systems and achieved over 90% mean accuracy while others were more challenging.
We report the results of our system on the Metaphor Detection Shared Task at the Second Workshop on Figurative Language Processing 2020. Our model is an ensemble, utilising contextualised and static distributional semantic representations, along with word-type concreteness ratings. Using these features, it predicts word metaphoricity with a deep multi-layer perceptron. We are able to best the state-of-the-art from the 2018 Shared Task by an average of 8.0% F1, and finish fourth in both sub-tasks in which we participate.
The Universal Morphology (UniMorph) project is a collaborative effort providing broad-coverage instantiated normalized morphological paradigms for hundreds of diverse world languages. The project comprises two major thrusts: a language-independent feature schema for rich morphological annotation and a type-level resource of annotated data in diverse languages realizing that schema. We have implemented several improvements to the extraction pipeline which creates most of our data, so that it is both more complete and more correct. We have added 66 new languages, as well as new parts of speech for 12 languages. We have also amended the schema in several ways. Finally, we present three new community tools: two to validate data for resource creators, and one to make morphological data available from the command line. UniMorph is based at the Center for Language and Speech Processing (CLSP) at Johns Hopkins University in Baltimore, Maryland. This paper details advances made to the schema, tooling, and dissemination of project resources since the UniMorph 2.0 release described at LREC 2018.
The performance of neural machine translation systems is commonly evaluated in terms of BLEU. However, due to its reliance on target language properties and generation, the BLEU metric does not allow an assessment of which translation directions are more difficult to model. In this paper, we propose cross-mutual information (XMI): an asymmetric information-theoretic metric of machine translation difficulty that exploits the probabilistic nature of most neural machine translation models. XMI allows us to better evaluate the difficulty of translating text into the target language while controlling for the difficulty of the target-side generation component independent of the translation task. We then present the first systematic and controlled study of cross-lingual translation difficulties using modern neural translation systems. Code for replicating our experiments is available online at https://github.com/e-bug/nmt-difficulty.
A major hurdle in data-driven research on typology is having sufficient data in many languages to draw meaningful conclusions. We present VoxClamantis v1.0, the first large-scale corpus for phonetic typology, with aligned segments and estimated phoneme-level labels in 690 readings spanning 635 languages, along with acoustic-phonetic measures of vowels and sibilants. Access to such data can greatly facilitate investigation of phonetic typology at a large scale and across many languages. However, it is non-trivial and computationally intensive to obtain such alignments for hundreds of languages, many of which have few to no resources presently available. We describe the methodology to create our corpus, discuss caveats with current methods and their impact on the utility of this data, and illustrate possible research directions through a series of case studies on the 48 highest-quality readings. Our corpus and scripts are publicly available for non-commercial use at https://voxclamantisproject.github.io.
The success of neural networks on a diverse set of NLP tasks has led researchers to question how much these networks actually “know” about natural language. Probes are a natural way of assessing this. When probing, a researcher chooses a linguistic task and trains a supervised model to predict annotations in that linguistic task from the network’s learned representations. If the probe does well, the researcher may conclude that the representations encode knowledge related to the task. A commonly held belief is that using simpler models as probes is better; the logic is that simpler models will identify linguistic structure, but not learn the task itself. We propose an information-theoretic operationalization of probing as estimating mutual information that contradicts this received wisdom: one should always select the highest performing probe one can, even if it is more complex, since it will result in a tighter estimate, and thus reveal more of the linguistic information inherent in the representation. The experimental portion of our paper focuses on empirically estimating the mutual information between a linguistic property and BERT, comparing these estimates to several baselines. We evaluate on a set of ten typologically diverse languages often underrepresented in NLP research—plus English—totalling eleven languages. Our implementation is available in https://github.com/rycolab/info-theoretic-probing.
The noun lexica of many natural languages are divided into several declension classes with characteristic morphological properties. Class membership is far from deterministic, but the phonological form of a noun and/or its meaning can often provide imperfect clues. Here, we investigate the strength of those clues. More specifically, we operationalize this by measuring how much information, in bits, we can glean about declension class from knowing the form and/or meaning of nouns. We know that form and meaning are often also indicative of grammatical gender—which, as we quantitatively verify, can itself share information with declension class—so we also control for gender. We find for two Indo-European languages (Czech and German) that form and meaning respectively share significant amounts of information with class (and contribute additional information above and beyond gender). The three-way interaction between class, form, and meaning (given gender) is also significant. Our study is important for two reasons: First, we introduce a new method that provides additional quantitative support for a classic linguistic finding that form and meaning are relevant for the classification of nouns into declensions. Secondly, we show not only that individual declensions classes vary in the strength of their clues within a language, but also that these variations themselves vary across languages.
Prior work has explored directly regularizing the output distributions of probabilistic models to alleviate peaky (i.e. over-confident) predictions, a common sign of overfitting. This class of techniques, of which label smoothing is one, has a connection to entropy regularization. Despite the consistent success of label smoothing across architectures and data sets in language generation tasks, two problems remain open: (1) there is little understanding of the underlying effects entropy regularizers have on models, and (2) the full space of entropy regularization techniques is largely unexplored. We introduce a parametric family of entropy regularizers, which includes label smoothing as a special case, and use it to gain a better understanding of the relationship between the entropy of a model and its performance on language generation tasks. We also find that variance in model performance can be explained largely by the resulting entropy of the model. Lastly, we find that label smoothing provably does not allow for sparsity in an output distribution, an undesirable property for language generation models, and therefore advise the use of other entropy regularization methods in its place.
Measuring what linguistic information is encoded in neural models of language has become popular in NLP. Researchers approach this enterprise by training “probes”—supervised models designed to extract linguistic structure from another model’s output. One such probe is the structural probe (Hewitt and Manning, 2019), designed to quantify the extent to which syntactic information is encoded in contextualised word representations. The structural probe has a novel design, unattested in the parsing literature, the precise benefit of which is not immediately obvious. To explore whether syntactic probes would do better to make use of existing techniques, we compare the structural probe to a more traditional parser with an identical lightweight parameterisation. The parser outperforms structural probe on UUAS in seven of nine analysed languages, often by a substantial amount (e.g. by 11.1 points in English). Under a second less common metric, however, there is the opposite trend—the structural probe outperforms the parser. This begs the question: which metric should we prefer?
This work treats the paradigm discovery problem (PDP), the task of learning an inflectional morphological system from unannotated sentences. We formalize the PDP and develop evaluation metrics for judging systems. Using currently available resources, we construct datasets for the task. We also devise a heuristic benchmark for the PDP and report empirical results on five diverse languages. Our benchmark system first makes use of word embeddings and string similarity to cluster forms by cell and by paradigm. Then, we bootstrap a neural transducer on top of the clustered data to predict words to realize the empty paradigm slots. An error analysis of our system suggests clustering by cell across different inflection classes is the most pressing challenge for future work.
We propose a novel morphologically aware probability model for bilingual lexicon induction, which jointly models lexeme translation and inflectional morphology in a structured way. Our model exploits the basic linguistic intuition that the lexeme is the key lexical unit of meaning, while inflectional morphology provides additional syntactic information. This approach leads to substantial performance improvements—19% average improvement in accuracy across 6 language pairs over the state of the art in the supervised setting and 16% in the weakly supervised setting. As another contribution, we highlight issues associated with modern BLI that stem from ignoring inflectional morphology, and propose three suggestions for improving the task.
Most modern NLP systems make use of pre-trained contextual representations that attain astonishingly high performance on a variety of tasks. Such high performance should not be possible unless some form of linguistic structure inheres in these representations, and a wealth of research has sprung up on probing for it. In this paper, we draw a distinction between intrinsic probing, which examines how linguistic information is structured within a representation, and the extrinsic probing popular in prior work, which only argues for the presence of such information by showing that it can be successfully extracted. To enable intrinsic probing, we propose a novel framework based on a decomposable multivariate Gaussian probe that allows us to determine whether the linguistic information in word embeddings is dispersed or focal. We then probe fastText and BERT for various morphosyntactic attributes across 36 languages. We find that most attributes are reliably encoded by only a few neurons, with fastText concentrating its linguistic structure more than BERT.
Quite surprisingly, exact maximum a posteriori (MAP) decoding of neural language generators frequently leads to low-quality results. Rather, most state-of-the-art results on language generation tasks are attained using beam search despite its overwhelmingly high search error rate. This implies that the MAP objective alone does not express the properties we desire in text, which merits the question: if beam search is the answer, what was the question? We frame beam search as the exact solution to a different decoding objective in order to gain insights into why high probability under a model alone may not indicate adequacy. We find that beam search enforces uniform information density in text, a property motivated by cognitive science. We suggest a set of decoding objectives that explicitly enforce this property and find that exact decoding with these objectives alleviates the problems encountered when decoding poorly calibrated language generation models. Additionally, we analyze the text produced using various decoding strategies and see that, in our neural machine translation experiments, the extent to which this property is adhered to strongly correlates with BLEU.
Bolukbasi et al. (2016) presents one of the first gender bias mitigation techniques for word embeddings. Their method takes pre-trained word embeddings as input and attempts to isolate a linear subspace that captures most of the gender bias in the embeddings. As judged by an analogical evaluation task, their method virtually eliminates gender bias in the embeddings. However, an implicit and untested assumption of their method is that the bias subspace is actually linear. In this work, we generalize their method to a kernelized, non-linear version. We take inspiration from kernel principal component analysis and derive a non-linear bias isolation technique. We discuss and overcome some of the practical drawbacks of our method for non-linear gender bias mitigation in word embeddings and analyze empirically whether the bias subspace is actually linear. Our analysis shows that gender bias is in fact well captured by a linear subspace, justifying the assumption of Bolukbasi et al. (2016).
The question of how to probe contextual word representations in a way that is principled and useful has seen significant recent attention. In our contribution to this discussion, we argue, first, for a probe metric that reflects the trade-off between probe complexity and performance: the Pareto hypervolume. To measure complexity, we present a number of parametric and non-parametric metrics. Our experiments with such metrics show that probe’s performance curves often fail to align with widely accepted rankings between language representations (with, e.g., non-contextual representations outperforming contextual ones). These results lead us to argue, second, that common simplistic probe tasks such as POS labeling and dependency arc labeling, are inadequate to evaluate the properties encoded in contextual word representations. We propose full dependency parsing as an example probe task, and demonstrate it with the Pareto hypervolume. In support of our arguments, the results of this illustrative experiment conform closer to accepted rankings among contextual word representations.
Lexical ambiguity is widespread in language, allowing for the reuse of economical word forms and therefore making language more efficient. If ambiguous words cannot be disambiguated from context, however, this gain in efficiency might make language less clear—resulting in frequent miscommunication. For a language to be clear and efficiently encoded, we posit that the lexical ambiguity of a word type should correlate with how much information context provides about it, on average. To investigate whether this is the case, we operationalise the lexical ambiguity of a word as the entropy of meanings it can take, and provide two ways to estimate this—one which requires human annotation (using WordNet), and one which does not (using BERT), making it readily applicable to a large number of languages. We validate these measures by showing that, on six high-resource languages, there are significant Pearson correlations between our BERT-based estimate of ambiguity and the number of synonyms a word has in WordNet (e.g. 𝜌 = 0.40 in English). We then test our main hypothesis—that a word’s lexical ambiguity should negatively correlate with its contextual uncertainty—and find significant correlations on all 18 typologically diverse languages we analyse. This suggests that, in the presence of ambiguity, speakers compensate by making contexts more informative.
Across languages, multiple consecutive adjectives modifying a noun (e.g. “the big red dog”) follow certain unmarked ordering rules. While explanatory accounts have been put forward, much of the work done in this area has relied primarily on the intuitive judgment of native speakers, rather than on corpus data. We present the first purely corpus-driven model of multi-lingual adjective ordering in the form of a latent-variable model that can accurately order adjectives across 24 different languages, even when the training and testing languages are different. We utilize this novel statistical model to provide strong converging evidence for the existence of universal, cross-linguistic, hierarchical adjective ordering tendencies.
The connection between dependency trees and spanning trees is exploited by the NLP community to train and to decode graph-based dependency parsers. However, the NLP literature has missed an important difference between the two structures: only one edge may emanate from the root in a dependency tree. We analyzed the output of state-of-the-art parsers on many languages from the Universal Dependency Treebank: although these parsers are often able to learn that trees which violate the constraint should be assigned lower probabilities, their ability to do so unsurprisingly de-grades as the size of the training set decreases. In fact, the worst constraint-violation rate we observe is 24%. Prior work has proposed an inefficient algorithm to enforce the constraint, which adds a factor of n to the decoding runtime. We adapt an algorithm due to Gabow and Tarjan (1984) to dependency parsing, which satisfies the constraint without compromising the original runtime.
A grammatical gender system divides a lexicon into a small number of relatively fixed grammatical categories. How similar are these gender systems across languages? To quantify the similarity, we define gender systems extensionally, thereby reducing the problem of comparisons between languages’ gender systems to cluster evaluation. We borrow a rich inventory of statistical tools for cluster evaluation from the field of community detection (Driver and Kroeber, 1932; Cattell, 1945), that enable us to craft novel information theoretic metrics for measuring similarity between gender systems. We first validate our metrics, then use them to measure gender system similarity in 20 languages. We then ask whether our gender system similarities alone are sufficient to reconstruct historical relationships between languages. Towards this end, we make phylogenetic predictions on the popular, but thorny, problem from historical linguistics of inducing a phylogenetic tree over extant Indo-European languages. Of particular interest, languages on the same branch of our phylogenetic tree are notably similar, whereas languages from separate branches are no more similar than chance.
We present methods for calculating a measure of phonotactic complexity—bits per phoneme— that permits a straightforward cross-linguistic comparison. When given a word, represented as a sequence of phonemic segments such as symbols in the international phonetic alphabet, and a statistical model trained on a sample of word types from the language, we can approximately measure bits per phoneme using the negative log-probability of that word under the model. This simple measure allows us to compare the entropy across languages, giving insight into how complex a language’s phonotactics is. Using a collection of 1016 basic concept words across 106 languages, we demonstrate a very strong negative correlation of − 0.74 between bits per phoneme and the average length of words.
Decoding for many NLP tasks requires an effective heuristic algorithm for approximating exact search because the problem of searching the full output space is often intractable, or impractical in many settings. The default algorithm for this job is beam search—a pruned version of breadth-first search. Quite surprisingly, beam search often returns better results than exact inference due to beneficial search bias for NLP tasks. In this work, we show that the standard implementation of beam search can be made up to 10x faster in practice. Our method assumes that the scoring function is monotonic in the sequence length, which allows us to safely prune hypotheses that cannot be in the final set of hypotheses early on. We devise effective monotonic approximations to popular nonmonontic scoring functions, including length normalization and mutual information decoding. Lastly, we propose a memory-reduced variant of best-first beam search, which has a similar beneficial search bias in terms of downstream performance, but runs in a fraction of the time.
We conduct a manual error analysis of the CoNLL-SIGMORPHON Shared Task on Morphological Reinflection. This task involves natural language generation: systems are given a word in citation form (e.g., hug) and asked to produce the corresponding inflected form (e.g., the simple past hugged). We propose an error taxonomy and use it to annotate errors made by the top two systems across twelve languages. Many of the observed errors are related to inflectional patterns sensitive to inherent linguistic properties such as animacy or affect; many others are failures to predict truly unpredictable inflectional behaviors. We also find nearly one quarter of the residual “errors” reflect errors in the gold data.
In this work, we propose the use of phone-level language models to estimate phonotactic complexity—measured in bits per phoneme—which makes cross-linguistic comparison straightforward. We compare the entropy across languages using this simple measure, gaining insight on how complex different language’s phonotactics are. Finally, we show a very strong negative correlation between phonotactic complexity and the average length of words—Spearman rho=-0.744—when analysing a collection of 106 languages with 1016 basic concepts each.
The SIGMORPHON 2019 shared task on cross-lingual transfer and contextual analysis in morphology examined transfer learning of inflection between 100 language pairs, as well as contextual lemmatization and morphosyntactic description in 66 languages. The first task evolves past years’ inflection tasks by examining transfer of morphological inflection knowledge from a high-resource language to a low-resource language. This year also presents a new second challenge on lemmatization and morphological feature analysis in context. All submissions featured a neural component and built on either this year’s strong baselines or highly ranked systems from previous years’ shared tasks. Every participating team improved in accuracy over the baselines for the inflection task (though not Levenshtein distance), and every team in the contextual analysis task improved on both state-of-the-art neural and non-neural baselines.
Human translators routinely have to translate rare inflections of words – due to the Zipfian distribution of words in a language. When translating from Spanish, a good translator would have no problem identifying the proper translation of a statistically rare inflection such as habláramos. Note the lexeme itself, hablar, is relatively common. In this work, we investigate whether state-of-the-art bilingual lexicon inducers are capable of learning this kind of generalization. We introduce 40 morphologically complete dictionaries in 10 languages and evaluate three of the best performing models on the task of translation of less frequent morphological forms. We demonstrate that the performance of state-of-the-art models drops considerably when evaluated on infrequent morphological inflections and then show that adding a simple morphological constraint at training time improves the performance, proving that the bilingual lexicon inducers can benefit from better encoding of morphology.
Can we construct a neural language model which is inductively biased towards learning human language? Motivated by this question, we aim at constructing an informative prior for held-out languages on the task of character-level, open-vocabulary language modelling. We obtain this prior as the posterior over network weights conditioned on the data from a sample of training languages, which is approximated through Laplace’s method. Based on a large and diverse sample of languages, the use of our prior outperforms baseline models with an uninformative prior in both zero-shot and few-shot settings, showing that the prior is imbued with universal linguistic knowledge. Moreover, we harness broad language-specific information available for most languages of the world, i.e., features from typological databases, as distant supervision for held-out languages. We explore several language modelling conditioning techniques, including concatenation and meta-networks for parameter generation. They appear beneficial in the few-shot setting, but ineffective in the zero-shot setting. Since the paucity of even plain digital text affects the majority of the world’s languages, we hope that these insights will broaden the scope of applications for language technology.
This paper treats gender bias latent in word embeddings. Previous mitigation attempts rely on the operationalisation of gender bias as a projection over a linear subspace. An alternative approach is Counterfactual Data Augmentation (CDA), in which a corpus is duplicated and augmented to remove bias, e.g. by swapping all inherently-gendered words in the copy. We perform an empirical comparison of these approaches on the English Gigaword and Wikipedia, and find that whilst both successfully reduce direct bias and perform well in tasks which quantify embedding quality, CDA variants outperform projection-based methods at the task of drawing non-biased gender analogies by an average of 19% across both corpora. We propose two improvements to CDA: Counterfactual Data Substitution (CDS), a variant of CDA in which potentially biased text is randomly substituted to avoid duplication, and the Names Intervention, a novel name-pairing technique that vastly increases the number of words being treated. CDA/S with the Names Intervention is the only approach which is able to mitigate indirect gender bias: following debiasing, previously biased words are significantly less clustered according to gender (cluster purity is reduced by 49%), thus improving on the state-of-the-art for bias mitigation.
Recent studies have shown that word embeddings exhibit gender bias inherited from the training corpora. However, most studies to date have focused on quantifying and mitigating such bias only in English. These analyses cannot be directly extended to languages that exhibit morphological agreement on gender, such as Spanish and French. In this paper, we propose new metrics for evaluating gender bias in word embeddings of these languages and further demonstrate evidence of gender bias in bilingual embeddings which align these languages with English. Finally, we extend an existing approach to mitigate gender bias in word embedding of these languages under both monolingual and bilingual settings. Experiments on modified Word Embedding Association Test, word similarity, word translation, and word pair translation tasks show that the proposed approaches can effectively reduce the gender bias while preserving the utility of the original embeddings.
Many of the world’s languages employ grammatical gender on the lexeme. For instance, in Spanish, house “casa” is feminine, whereas the word for paper “papel” is masculine. To a speaker of a genderless language, this categorization seems to exist with neither rhyme nor reason. But, is the association of nouns to gender classes truly arbitrary? In this work, we present the first large-scale investigation of the arbitrariness of gender assignment that uses canonical correlation analysis as a method for correlating the gender of inanimate nouns with their lexical semantic meaning. We find that the gender systems of 18 languages exhibit a significant correlation with an externally grounded definition of lexical semantics.
In this paper, we quantify, analyze and mitigate gender bias exhibited in ELMo’s contextualized word vectors. First, we conduct several intrinsic analyses and find that (1) training data for ELMo contains significantly more male than female entities, (2) the trained ELMo embeddings systematically encode gender information and (3) ELMo unequally encodes gender information about male and female entities. Then, we show that a state-of-the-art coreference system that depends on ELMo inherits its bias and demonstrates significant bias on the WinoBias probing corpus. Finally, we explore two methods to mitigate such gender bias and show that the bias demonstrated on WinoBias can be eliminated.
When assigning quantitative labels to a dataset, different methodologies may rely on different scales. In particular, when assigning polarities to words in a sentiment lexicon, annotators may use binary, categorical, or continuous labels. Naturally, it is of interest to unify these labels from disparate scales to both achieve maximal coverage over words and to create a single, more robust sentiment lexicon while retaining scale coherence. We introduce a generative model of sentiment lexica to combine disparate scales into a common latent representation. We realize this model with a novel multi-view variational autoencoder (VAE), called SentiVAE. We evaluate our approach via a downstream text classification task involving nine English-Language sentiment analysis datasets; our representation outperforms six individual sentiment lexica, as well as a straightforward combination thereof.
English verbs have multiple forms. For instance, talk may also appear as talks, talked or talking, depending on the context. The NLP task of lemmatization seeks to map these diverse forms back to a canonical one, known as the lemma. We present a simple joint neural model for lemmatization and morphological tagging that achieves state-of-the-art results on 20 languages from the Universal Dependencies corpora. Our paper describes the model in addition to training and decoding procedures. Error analysis indicates that joint morphological tagging and lemmatization is especially helpful in low-resource lemmatization and languages that display a larger degree of morphological complexity.
In the principles-and-parameters framework, the structural features of languages depend on parameters that may be toggled on or off, with a single parameter often dictating the status of multiple features. The implied covariance between features inspires our probabilisation of this line of linguistic inquiry—we develop a generative model of language based on exponential-family matrix factorisation. By modelling all languages and features within the same architecture, we show how structural similarities between languages can be exploited to predict typological features with near-perfect accuracy, outperforming several baselines on the task of predicting held-out features. Furthermore, we show that language embeddings pre-trained on monolingual text allow for generalisation to unobserved languages. This finding has clear practical and also theoretical implications: the results confirm what linguists have hypothesised, i.e. that there are significant correlations between typological features and languages.
Critical to natural language generation is the production of correctly inflected text. In this paper, we isolate the task of predicting a fully inflected sentence from its partially lemmatized version. Unlike traditional morphological inflection or surface realization, our task input does not provide “gold” tags that specify what morphological features to realize on each lemmatized word; rather, such features must be inferred from sentential context. We develop a neural hybrid graphical model that explicitly reconstructs morphological features before predicting the inflected forms, and compare this to a system that directly predicts the inflected forms without relying on any morphological annotation. We experiment on several typologically diverse languages from the Universal Dependencies treebanks, showing the utility of incorporating linguistically-motivated latent variables into NLP models.
While idiosyncrasies of the Chinese classifier system have been a richly studied topic among linguists (Adams and Conklin, 1973; Erbaugh, 1986; Lakoff, 1986), not much work has been done to quantify them with statistical methods. In this paper, we introduce an information-theoretic approach to measuring idiosyncrasy; we examine how much the uncertainty in Mandarin Chinese classifiers can be reduced by knowing semantic information about the nouns that the classifiers modify. Using the empirical distribution of classifiers from the parsed Chinese Gigaword corpus (Graff et al., 2005), we compute the mutual information (in bits) between the distribution over classifiers and distributions over other linguistic quantities. We investigate whether semantic classes of nouns and adjectives differ in how much they reduce uncertainty in classifier choice, and find that it is not fully idiosyncratic; while there are no obvious trends for the majority of semantic classes, shape nouns reduce uncertainty in classifier choice the most.
Many common character-level, string-to-string transduction tasks, e.g., grapheme-to-phoneme conversion and morphological inflection, consist almost exclusively of monotonic transduction. Neural sequence-to-sequence models with soft attention, non-monotonic models, outperform popular monotonic models. In this work, we ask the following question: Is monotonicity really a helpful inductive bias in these tasks? We develop a hard attention sequence-to-sequence model that enforces strict monotonicity and learns alignment jointly. With the help of dynamic programming, we are able to compute the exact marginalization over all alignments. Our models achieve state-of-the-art performance on morphological inflection. Furthermore, we find strong performance on two other character-level transduction tasks. Code is available at https://github.com/shijie-wu/neural-transducer.
Gender stereotypes are manifest in most of the world’s languages and are consequently propagated or amplified by NLP systems. Although research has focused on mitigating gender stereotypes in English, the approaches that are commonly employed produce ungrammatical sentences in morphologically rich languages. We present a novel approach for converting between masculine-inflected and feminine-inflected sentences in such languages. For Spanish and Hebrew, our approach achieves F1 scores of 82% and 73% at the level of tags and accuracies of 90% and 87% at the level of forms. By evaluating our approach using four different languages, we show that, on average, it reduces gender stereotyping by a factor of 2.5 without any sacrifice to grammaticality.
Studying the ways in which language is gendered has long been an area of interest in sociolinguistics. Studies have explored, for example, the speech of male and female characters in film and the language used to describe male and female politicians. In this paper, we aim not to merely study this phenomenon qualitatively, but instead to quantify the degree to which the language used to describe men and women is different and, moreover, different in a positive or negative way. To that end, we introduce a generative latent-variable model that jointly represents adjective (or verb) choice, with its sentiment, given the natural gender of a head (or dependent) noun. We find that there are significant differences between descriptions of male and female nouns and that these differences align with common gender stereotypes: Positive adjectives used to describe women are more often related to their bodies than adjectives used to describe men.
A longstanding debate in semiotics centers on the relationship between linguistic signs and their corresponding semantics: is there an arbitrary relationship between a word form and its meaning, or does some systematic phenomenon pervade? For instance, does the character bigram ‘gl’ have any systematic relationship to the meaning of words like ‘glisten’, ‘gleam’ and ‘glow’? In this work, we offer a holistic quantification of the systematicity of the sign using mutual information and recurrent neural networks. We employ these in a data-driven and massively multilingual approach to the question, examining 106 languages. We find a statistically significant reduction in entropy when modeling a word form conditioned on its semantic representation. Encouragingly, we also recover well-attested English examples of systematic affixes. We conclude with the meta-point: Our approximate effect size (measured in bits) is quite small—despite some amount of systematicity between form and meaning, an arbitrary relationship and its resulting benefits dominate human language.
The study of linguistic typology is rooted in the implications we find between linguistic features, such as the fact that languages with object-verb word ordering tend to have postpositions. Uncovering such implications typically amounts to time-consuming manual processing by trained and experienced linguists, which potentially leaves key linguistic universals unexplored. In this paper, we present a computational model which successfully identifies known universals, including Greenberg universals, but also uncovers new ones, worthy of further linguistic investigation. Our approach outperforms baselines previously used for this problem, as well as a strong baseline from knowledge base population.
Embedding a clause inside another (“the girl [who likes cars [that run fast]] has arrived”) is a fundamental resource that has been argued to be a key driver of linguistic expressiveness. As such, it plays a central role in fundamental debates on what makes human language unique, and how they might have evolved. Empirical evidence on the prevalence and the limits of embeddings has however been based on either laboratory setups or corpus data of relatively limited size. We introduce here a collection of large, dependency-parsed written corpora in 17 languages, that allow us, for the first time, to capture clausal embedding through dependency graphs and assess their distribution. Our results indicate that there is no evidence for hard constraints on embedding depth: the tail of depth distributions is heavy. Moreover, although deeply embedded clauses tend to be shorter, suggesting processing load issues, complex sentences with many embeddings do not display a bias towards less deep embeddings. Taken together, the results suggest that deep embeddings are not disfavoured in written language. More generally, our study illustrates how resources and methods from latest-generation big-data NLP can provide new perspectives on fundamental questions in theoretical linguistics.
How language-agnostic are current state-of-the-art NLP tools? Are there some types of language that are easier to model with current methods? In prior work (Cotterell et al., 2018) we attempted to address this question for language modeling, and observed that recurrent neural network language models do not perform equally well over all the high-resource European languages found in the Europarl corpus. We speculated that inflectional morphology may be the primary culprit for the discrepancy. In this paper, we extend these earlier experiments to cover 69 languages from 13 language families using a multilingual Bible corpus. Methodologically, we introduce a new paired-sample multiplicative mixed-effects model to obtain language difficulty coefficients from at-least-pairwise parallel corpora. In other words, the model is aware of inter-sentence variation and can handle missing data. Exploiting this model, we show that “translationese” is not any easier to model than natively written language in a fair comparison. Trying to answer the question of what features difficult languages have in common, we try and fail to reproduce our earlier (Cotterell et al., 2018) observation about morphological complexity and instead reveal far simpler statistics of the data that seem to drive complexity in a much larger sample.
We present a study of morphological irregularity. Following recent work, we define an information-theoretic measure of irregularity based on the predictability of forms in a language. Using a neural transduction model, we estimate this quantity for the forms in 28 languages. We first present several validatory and exploratory analyses of irregularity. We then show that our analyses provide evidence for a correlation between irregularity and frequency: higher frequency items are more likely to be irregular and irregular items are more likely be highly frequent. To our knowledge, this result is the first of its breadth and confirms longstanding proposals from the linguistics literature. The correlation is more robust when aggregated at the level of whole paradigms—providing support for models of linguistic structure in which inflected forms are unified by abstract underlying stems or lexemes.
We quantify the linguistic complexity of different languages’ morphological systems. We verify that there is a statistically significant empirical trade-off between paradigm size and irregularity: A language’s inflectional paradigms may be either large in size or highly irregular, but never both. We define a new measure of paradigm irregularity based on the conditional entropy of the surface realization of a paradigm— how hard it is to jointly predict all the word forms in a paradigm from the lemma. We estimate irregularity by training a predictive model. Our measurements are taken on large morphological paradigms from 36 typologically diverse languages.
Much like sentences are composed of words, words themselves are composed of smaller units. For example, the English word questionably can be analyzed as question+able+ly. However, this structural decomposition of the word does not directly give us a semantic representation of the word’s meaning. Since morphology obeys the principle of compositionality, the semantics of the word can be systematically derived from the meaning of its parts. In this work, we propose a novel probabilistic model of word formation that captures both the analysis of a word w into its constituent segments and the synthesis of the meaning of w from the meanings of those segments. Our model jointly learns to segment words into morphemes and compose distributional semantic vectors of those morphemes. We experiment with the model on English CELEX data and German DErivBase (Zeller et al., 2013) data. We show that jointly modeling semantics increases both segmentation accuracy and morpheme F1 by between 3% and 5%. Additionally, we investigate different models of vector composition, showing that recurrent neural networks yield an improvement over simple additive models. Finally, we study the degree to which the representations correspond to a linguist’s notion of morphological productivity.
Can advances in NLP help advance cognitive modeling? We examine the role of artificial neural networks, the current state of the art in many common NLP tasks, by returning to a classic case study. In 1986, Rumelhart and McClelland famously introduced a neural architecture that learned to transduce English verb stems to their past tense forms. Shortly thereafter in 1988, Pinker and Prince presented a comprehensive rebuttal of many of Rumelhart and McClelland’s claims. Much of the force of their attack centered on the empirical inadequacy of the Rumelhart and McClelland model. Today, however, that model is severely outmoded. We show that the Encoder-Decoder network architectures used in modern NLP systems obviate most of Pinker and Prince’s criticisms without requiring any simplification of the past tense mapping problem. We suggest that the empirical performance of modern networks warrants a reexamination of their utility in linguistic and cognitive modeling.
What makes some types of languages more probable than others? For instance, we know that almost all spoken languages contain the vowel phoneme /i/; why should that be? The field of linguistic typology seeks to answer these questions and, thereby, divine the mechanisms that underlie human language. In our work, we tackle the problem of vowel system typology, i.e., we propose a generative probability model of which vowels a language contains. In contrast to previous work, we work directly with the acoustic information—the first two formant values—rather than modeling discrete sets of symbols from the international phonetic alphabet. We develop a novel generative probability model and report results on over 200 languages.
For general modeling methods applied to diverse languages, a natural question is: how well should we expect our models to work on languages with differing typological profiles? In this work, we develop an evaluation framework for fair cross-linguistic comparison of language models, using translated text so that all models are asked to predict approximately the same information. We then conduct a study on 21 languages, demonstrating that in some languages, the textual expression of the information is harder to predict with both n-gram and LSTM language models. We show complex inflectional morphology to be a cause of performance differences among languages.
Lexical ambiguity makes it difficult to compute useful statistics of a corpus. A given word form might represent any of several morphological feature bundles. One can, however, use unsupervised learning (as in EM) to fit a model that probabilistically disambiguates word forms. We present such an approach, which employs a neural network to smoothly model a prior distribution over feature bundles (even rare ones). Although this basic model does not consider a token’s context, that very property allows it to operate on a simple list of unigram type counts, partitioning each count among different analyses of that unigram. We discuss evaluation metrics for this novel task and report results on 5 languages.
Most recent approaches to bilingual dictionary induction find a linear alignment between the word vector spaces of two languages. We show that projecting the two languages onto a third, latent space, rather than directly onto each other, while equivalent in terms of expressivity, makes it easier to learn approximate alignments. Our modified approach also allows for supporting languages to be included in the alignment process, to obtain an even better performance in low resource settings.
We introduce a novel discriminative latent-variable model for the task of bilingual lexicon induction. Our model combines the bipartite matching dictionary prior of Haghighi et al. (2008) with a state-of-the-art embedding-based approach. To train the model, we derive an efficient Viterbi EM algorithm. We provide empirical improvements on six language pairs under two metrics and show that the prior theoretically and empirically helps to mitigate the hubness problem. We also demonstrate how previous work may be viewed as a similarly fashioned latent-variable model, albeit with a different prior.
Character-level string-to-string transduction is an important component of various NLP tasks. The goal is to map an input string to an output string, where the strings may be of different lengths and have characters taken from different alphabets. Recent approaches have used sequence-to-sequence models with an attention mechanism to learn which parts of the input string the model should focus on during the generation of the output string. Both soft attention and hard monotonic attention have been used, but hard non-monotonic attention has only been used in other sequence modeling tasks and has required a stochastic approximation to compute the gradient. In this work, we introduce an exact, polynomial-time algorithm for marginalizing over the exponential number of non-monotonic alignments between two strings, showing that hard attention models can be viewed as neural reparameterizations of the classical IBM Model 1. We compare soft and hard non-monotonic attention experimentally and find that the exact algorithm significantly improves performance over the stochastic approximation and outperforms soft attention.
Statistical morphological inflectors are typically trained on fully supervised, type-level data. One remaining open research question is the following: How can we effectively exploit raw, token-level data to improve their performance? To this end, we introduce a novel generative latent-variable model for the semi-supervised learning of inflection generation. To enable posterior inference over the latent variables, we derive an efficient variational inference procedure based on the wake-sleep algorithm. We experiment on 23 languages, using the Universal Dependencies corpora in a simulated low-resource setting, and find improvements of over 10% absolute accuracy in some cases.
The Universal Dependencies (UD) and Universal Morphology (UniMorph) projects each present schemata for annotating the morphosyntactic details of language. Each project also provides corpora of annotated text in many languages—UD at the token level and UniMorph at the type level. As each corpus is built by different annotators, language-specific decisions hinder the goal of universal schemata. With compatibility of tags, each project’s annotations could be used to validate the other’s. Additionally, the availability of both type- and token-level resources would be a boon to tasks such as parsing and homograph disambiguation. To ease this interoperability, we present a deterministic mapping from Universal Dependencies v2 features into the UniMorph schema. We validate our approach by lookup in the UniMorph corpora and find a macro-average of 64.13% recall. We also note incompatibilities due to paucity of data on either side. Finally, we present a critical evaluation of the foundations, strengths, and weaknesses of the two annotation projects.
We explore the task of multi-source morphological reinflection, which generalizes the standard, single-source version. The input consists of (i) a target tag and (ii) multiple pairs of source form and source tag for a lemma. The motivation is that it is beneficial to have access to more than one source form since different source forms can provide complementary information, e.g., different stems. We further present a novel extension to the encoder-decoder recurrent neural architecture, consisting of multiple encoders, to better solve the task. We show that our new architecture outperforms single-source reinflection models and publish our dataset for multi-source morphological reinflection to facilitate future research.
A traditional claim in linguistics is that all human languages are equally expressive—able to convey the same wide range of meanings. Morphologically rich languages, such as Czech, rely on overt inflectional and derivational morphology to convey many semantic distinctions. Languages with comparatively limited morphology, such as English, should be able to accomplish the same using a combination of syntactic and contextual cues. We capitalize on this idea by training a tagger for English that uses syntactic features obtained by automatic parsing to recover complex morphological tags projected from Czech. The high accuracy of the resulting model provides quantitative confirmation of the underlying linguistic hypothesis of equal expressivity, and bodes well for future improvements in downstream HLT tasks including machine translation.
Derivational morphology is a fundamental and complex characteristic of language. In this paper we propose a new task of predicting the derivational form of a given base-form lemma that is appropriate for a given context. We present an encoder-decoder style neural network to produce a derived form character-by-character, based on its corresponding character-level representation of the base form and the context. We demonstrate that our model is able to generate valid context-sensitive derivations from known base forms, but is less accurate under lexicon agnostic setting.
The popular skip-gram model induces word embeddings by exploiting the signal from word-context coocurrence. We offer a new interpretation of skip-gram based on exponential family PCA-a form of matrix factorization to generalize the skip-gram model to tensor factorization. In turn, this lets us train embeddings through richer higher-order coocurrences, e.g., triples that include positional information (to incorporate syntax) or morphological information (to share parameters across related words). We experiment on 40 languages and show our model improves upon skip-gram.
The Dravidian languages are one of the most widely spoken language families in the world, yet there are very few annotated resources available to NLP researchers. To remedy this, we create DravMorph, a corpus annotated for morphological segmentation and part-of-speech. Additionally, we exploit novel features and higher-order models to set state-of-the-art results on these corpora on both tasks, beating techniques proposed in the literature by as much as 4 points in segmentation F1.
Many of the world’s languages contain an abundance of inflected forms for each lexeme. A critical task in processing such languages is predicting these inflected forms. We develop a novel statistical model for the problem, drawing on graphical modeling techniques and recent advances in deep learning. We derive a Metropolis-Hastings algorithm to jointly decode the model. Our Bayesian network draws inspiration from principal parts morphological analysis. We demonstrate improvements on 5 languages.
We study how different frame annotations complement one another when learning continuous lexical semantics. We learn the representations from a tensorized skip-gram model that consistently encodes syntactic-semantic content better, with multiple 10% gains over baselines.
Linguistic typology studies the range of structures present in human language. The main goal of the field is to discover which sets of possible phenomena are universal, and which are merely frequent. For example, all languages have vowels, while most—but not all—languages have an /u/ sound. In this paper we present the first probabilistic treatment of a basic question in phonological typology: What makes a natural vowel inventory? We introduce a series of deep stochastic point processes, and contrast them with previous computational, simulation-based approaches. We provide a comprehensive suite of experiments on over 200 distinct languages.
We present a novel cross-lingual transfer method for paradigm completion, the task of mapping a lemma to its inflected forms, using a neural encoder-decoder model, the state of the art for the monolingual task. We use labeled data from a high-resource language to increase performance on a low-resource language. In experiments on 21 language pairs from four different language families, we obtain up to 58% higher accuracy than without transfer and show that even zero-shot and one-shot learning are possible. We further find that the degree of language relatedness strongly influences the ability to transfer morphological knowledge.
The generation of complex derived word forms has been an overlooked problem in NLP; we fill this gap by applying neural sequence-to-sequence models to the task. We overview the theoretical motivation for a paradigmatic treatment of derivational morphology, and introduce the task of derivational paradigm completion as a parallel to inflectional paradigm completion. State-of-the-art neural models adapted from the inflection task are able to learn the range of derivation patterns, and outperform a non-neural baseline by 16.4%. However, due to semantic, historical, and lexical considerations involved in derivational morphology, future work will be needed to achieve performance parity with inflection-generating systems.
Even for common NLP tasks, sufficient supervision is not available in many languages – morphological tagging is no exception. In the work presented here, we explore a transfer learning scheme, whereby we train character-level recurrent neural taggers to predict morphological taggings for high-resource languages and low-resource languages together. Learning joint character representations among multiple related languages successfully enables knowledge transfer from the high-resource languages to the low-resource ones.
Low-resource named entity recognition is still an open problem in NLP. Most state-of-the-art systems require tens of thousands of annotated sentences in order to obtain high performance. However, for most of the world’s languages it is unfeasible to obtain such annotation. In this paper, we present a transfer learning scheme, whereby we train character-level neural CRFs to predict named entities for both high-resource languages and low-resource languages jointly. Learning character representations for multiple related languages allows knowledge transfer from the high-resource languages to the low-resource ones, improving F1 by up to 9.8 points.
The observed pronunciations or spellings of words are often explained as arising from the “underlying forms” of their morphemes. These forms are latent strings that linguists try to reconstruct by hand. We propose to reconstruct them automatically at scale, enabling generalization to new words. Given some surface word types of a concatenative language along with the abstract morpheme sequences that they express, we show how to recover consistent underlying forms for these morphemes, together with the (stochastic) phonology that maps each concatenation of underlying forms to a surface form. Our technique involves loopy belief propagation in a natural directed graphical model whose variables are unknown strings and whose conditional distributions are encoded as finite-state machines with trainable weights. We define training and evaluation paradigms for the task of surface word prediction, and report results on subsets of 7 languages.
This paper presents a multi-dialect, multi-genre, human annotated corpus of dialectal Arabic. We collected utterances in five Arabic dialects: Levantine, Gulf, Egyptian, Iraqi and Maghrebi. We scraped newspaper websites for user commentary and Twitter for two distinct types of dialectal content. To the best of the authors knowledge, this work is the most diverse corpus of dialectal Arabic in both the source of the content and the number of dialects. Every utterance in the corpus was human annotated on Amazons Mechanical Turk; this stands in contrast to Al-Sabbagh and Girju (2012) where only a small subset was human annotated in order to train a classifier to automatically annotate the remainder of the corpus. We provide a discussion of the methodology used for the annotation in addition to the performance of the individual workers. We extend the Arabic dialect identification task to the Iraqi and Maghrebi dialects and improve the results of Zaidan and Callison-Burch (2011a) on Levantine, Gulf and Egyptian.
Translation of the output of automatic speech recognition (ASR) systems, also known as speech translation, has received a lot of research interest recently. This is especially true for programs such as DARPA BOLT which focus on improving spontaneous human-human conversation across languages. However, this research is hindered by the dearth of datasets developed for this explicit purpose. For Egyptian Arabic-English, in particular, no parallel speechtranscription-translation dataset exists in the same domain. In order to support research in speech translation, we introduce the Callhome Egyptian Arabic-English Speech Translation Corpus. This supplements the existing LDC corpus with four reference translations for each utterance in the transcripts. The result is a three-way parallel dataset of Egyptian Arabic Speech, transcriptions and English translations.
