Redundancy and reduction: Speakers manage syntactic information density

By T. Florian Jaeger. Cognitive Psychology, 2010

702 citations recorded by the source corpus.

A principle of efficient language production based on information theoretic considerations is proposed: Uniform Information Density predicts that language production is affected by a preference to distribute information uniformly across the linguistic signal. This prediction is tested against data from syntactic reduction. A single multilevel logit model analysis of naturally distributed data from a corpus of spontaneous speech is used to assess the effect of information density on complementizer that-mentioning, while simultaneously evaluating the predictions of several influential alternative accounts: availability, ambiguity avoidance, and dependency processing accounts. Information density emerges as an important predictor of speakers’ preferences during production. As information is defined in terms of probabilities, it follows that production is probability-sensitive, in that speakers’ preferences are affected by the contextual probability of syntactic structures. The merits of a corpus-based approach to the study of language production are discussed as well.

Subfields: Cognitive Neuroscience, Cultural Studies, Developmental and Educational Psychology

View Paper: https://www.semanticscholar.org/paper/0ac50fcd3fe81fb7dd55da2ecbaff5de639f8f1f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Cultural Studies

In Developmental and Educational Psychology

Citations in the corpus

Word lengths are optimized for efficient communication (2011, background): “A tendency for this type of “ smoothing out ” peaks and dips of informativeness is known as uniform information density and has been observed in choices made during online language production (7 – 10).”
Investigating locality effects and surprisal in written English syntactic choice phenomena. (2016, unknown): “There is also parallel evidence from optional function words, which are likely to be omitted to shorten dependencies (Hawkins 2001; 2003, Jaeger 2006; 2010; 2011).”
What do we mean by prediction in language comprehension? (2015, background): “They might even have learned generative models for situations in which normal statistics completely break down, such as when participating in experiments (cf. Jaeger, 2010, p. 53) or when talking to someone believed to have a language deficit (Arnold, Kam, & Tanenhaus, 2007).”
Zipf’s word frequency law in natural language: A critical review and future directions (2014, background): “…be intrinsically linked to theories of what human language comprehenders find predictable (e.g. Demberg & Keller, 2008; A. Frank & Jaeger, 2008; Jaeger, 2010; Levy, 2008; Levy & Jaeger, 2007; Piantadosi, Tily, & Gibson, 2011; N. J. Smith & Levy, in press) and how much information is…”
Random effects structure for confirmatory hypothesis testing: Keep it maximal (2013, unknown): “Discussion of such techniques for complex datasets and their proper application of is beyond the scope of this paper (but see, e.g., Baayen, 2008; Jaeger, 2010).”
Why discourse affects speakers’ choice of referring expressions (2015, background): “These are the same predictions that have been discussed in the context of UID, but here the predictions are derived from a principled model of speakers who are trying to provide information to listeners.”
of construction probability on word durations during spontaneous incremental sentence production (2012, methodology): “On UID account, it is at this point that speakers accommodate in the acoustic signal the information density associated with the choice, such that a less probable (more informative) alternative would show an increased amount of signal (slower or more disfluent speech, or a higher prevalence of gestu”
Rapid Expectation Adaptation during Syntactic Comprehension (2013, background): “Finally, our findings, together with previous work on syntactic priming, pose a methodological question for behavioral work on language processing: if humans continuously adapt their expectations to the statistics of the linguistic environment, then adaptation should be operative in any behavioral e”
The Goldilocks Effect: Human Infants Allocate Attention to Visual Sequences That Are Neither Too Simple Nor Too Complex (2012, background): “Similar hypotheses about how adults allocate their limited resources in the language domain–for example, those supporting a uniform information principle [33–37]–may suggest that what we have observed in infants reflects a ubiquitous constraint across domains and developmental levels.”
What Do Recurrent Neural Network Grammars Learn About Syntax? (2016, background): “This finding is intuitive, as complementizer that is often optional (Jaeger, 2010), unlike prepositional words that might describe relative time and location.”
If Beam Search Is the Answer, What Was the Question? (2020, background): “Where speakers have a choice be-tween several variants to encode their message, they prefer the variant with more uniform information density (ceteris paribus)” (Jaeger, 2010).”
The communicative function of ambiguity in language. (2011, unknown): “Similarly, structural ambiguities that slow down human comprehension are extremely rare (Wasow & Arnold, 2003; Jaeger, 2006; Ferreira, 2008; Jaeger, 2010).”
Why reduce? Phonological neighborhood density and phonetic reduction in spontaneous speech (2012, background): “The relationship between neighborhood density and confusability conforms to many people’s 1 Several of these proposals (Aylett & Turk, 2004; Jaeger, 2010; Levy & Jaeger, 2007; van Son & Pols, 2003) relate the reduction of highlypredictable forms to the pacing of information density throughout…”
What residualizing predictors in regression analyses does (and what it does not do) (2014, methodology): “Such sign changes are what many researchers hope to avoid by residualizing (e.g., Jaeger, 2010), but Residualizing Predictors in Regression 11 Ambridge et al. conclude that residualizing in fact caused them: ...the residualized pre-emption predictor is working in the opposite direction to that…”