Volume 7, Issue s3 - Efficiency in human languages: corpus evidence for universal principles / edited by Natalia Levshina and Steven Moran

Over the last few years, there has been a growing interest in communicative efficiency. It has been argued that language users act efficiently, saving effort for processing and articulation, and that language structure and use reflect this tendency. The emergence of new corpus data has brought to life numerous studies on efficient language use in the lexicon, in morphosyntax, and in discourse and phonology in different languages. In this introductory paper, we discuss communicative efficiency in human languages, focusing on evidence of efficient language use found in multilingual corpora. The evidence suggests that efficiency is a universal feature of human language. We provide an overview of different manifestations of efficiency on different levels of language structure, and we discuss the major questions and findings so far, some of which are addressed for the first time in the contributions in this special collection.

In syntactic dependency trees, when arcs are drawn from syntactic heads to dependents, they rarely cross. Constraints on these crossing dependencies are critical for determining the syntactic properties of human language, because they define the position of natural language in formal language hierarchies. We study whether the apparent constraints on crossing syntactic dependencies in natural language might be explained by constraints on dependency lengths (the linear distance between heads and dependents). We compare real dependency trees from treebanks of 52 languages against baselines of random trees which are matched with the real trees in terms of their dependency lengths. We find that these baseline trees have many more crossing dependencies than real trees, indicating that a constraint on dependency lengths alone cannot explain the empirical rarity of crossing dependencies. However, we find evidence that a combined constraint on dependency length and the rate of crossing dependencies might be able to explain two of the most-studied formal restrictions on dependency trees: gap degree and well-nestedness.

Zipf’s Law of Abbreviation and Menzerath’s Law both make predictions about the length of linguistic units, based on corpus frequency and the length of the carrier unit. Each contributes to the efficiency of languages: for Zipf, units are more likely to be reduced when they are highly predictable, due to their frequency; for Menzerath, units are more likely to be reduced when there are more sub-units to contribute to the structural information of the carrier unit. However, it remains unclear how the two laws work together in determining unit length at a given level of linguistic structure. We examine this question regarding the length of morphemes in spoken corpora of nine typologically diverse languages drawn from the DoReCo corpus, showing that Zipf’s Law is a stronger predictor, but that the two laws interact with one another. We also explore how this is affected by specific typological characteristics, such as morphological complexity.

The introduction of new referents into discourse has traditionally been regarded as a major challenge to language processing, for which speakers deploy specific syntactic configurations, guided by the speaker’s assessment of the recipient’s state of mind (‘recipient design’). In this paper we probe these assumptions against discourse data from nine languages. We find little evidence for specialized syntactic configurations accommodating new referents; the only notable exception is the association of new reference with direct objects, suggests that linking new referents to already established discourse frames through a transitive construction is preferable to isolating them in an intransitive one. Where specific intransitive predicates are indeed found to host new referents, we find this to be motivated primarily by semantic considerations. Contrary to long-held assumptions, we conclude that the cognitive challenge of referent introduction is only weakly reflected in morphosyntax; instead, discourse production is most efficient when new referents are integrated seamlessly with content-driven demands of the narration.

Information theory can be used to assess how efficiently a message is transmitted on the basis of different symbolic systems. In this paper, I estimate the information-theoretic efficiency of written language for parallel text data in more than 1000 different languages, both on the level of characters and on the level of words as information encoding units. The main results show that (i) the median efficiency is ∼29% on the character level and ∼45% on the word level, (ii) efficiency on both levels is strongly correlated with each other and (iii) efficiency tends to be higher for languages with more speakers.

The use of differential case marking of A and P has been explained in terms of efficiency (economy) and markedness. The present study tests predictions based on these accounts, using conditional probabilities of a particular feature given the syntactic role (cue availability), and conditional probabilities of a particular syntactic role given the feature in question (cue reliability). Cue availability serves as a measure of markedness, whereas cue reliability is central for the efficiency account. Similar to reverse engineering, we determine which of the probabilistic measures could have been responsible for the recurrent cross-linguistic patterns described in the literature. The probabilities are estimated from spontaneous informal dialogues in English and Russian (Indo-European), Lao (Tai-Kadai), N||ng (Tuu) and Ruuli (Bantu). The analyses, which involve a series of mixed-effects Poisson models, clearly demonstrate that cue reliability matches the observed cross-linguistic patterns better than cue availability. Thus, the results support the efficiency account of differential marking.

Since (Zipf, George Kingsley. 1935. The psychobiology of language: An introduction to dynamic philology . Cambridge, MA: MIT Press; Zipf, George Kingsley. 1949. Human behavior and the principle of least effort. Journal of Consulting Psychology 13(3)), it has been known that more frequent lexical items tend to be shorter than less frequent ones, and this association between the length of an expression and its frequency has been applied to various grammatical patterns (syntactic, morphological, and phonological) and related to predictability or expectedness in the typological literature. However, the exact interactions of frequency and expectedness, their effect on shortening, and the mechanisms involved, are still not well understood. This paper proposes the Form-Expectedness Correspondence Hypothesis ( fech ), taking into account not only the frequency of expressions but their overall structure and distribution, and explores the fech in the domain of nominal inflection from a quantitative perspective.