Papers
Kleinschmidt, D.F., Fine, A.B., and Jaeger, T.F. 2012. A belief-updating model of adaptation and cue combination in syntactic comprehension. The 34th Annual Meeting of the Cognitive Science Society (CogSci12). Sapporo, Japan. July, 2012.
Feel free to cite. For page numbers, pls see the CogSci Proceedings webpage.
We develop and evaluate a preliminary belief-updating model which links intermediate-term (i.e., over several days) syntactic adaptation to the joint statistics of syntactic structures and lexical cues to those structures. This model shows how subjects differentially depend on different cues to syntactic structure following changes in the reliability of those cues, as shown by Fine and Jaeger (2011). By relating syntactic adaptation and cue combination to rational inference under uncertainty, this work links learning and adaptation in sentence processing with adaptation in speech perception and non-linguistic domains.
Keywords: sentence processing, adaptation, Bayesian modeling
13 views
Seen by:
Download (.pdf) Comment on “Phonemic Diversity Supports a Serial Founder Effect Model of Language Expansion from Africa”
This is the author's version of the work. It is posted here by permission of the AAAS for personal use, not for redistribution. The definitive version was published in Science 335, 2 MARCH 2012, doi:10.1126/science.1215107
Atkinson (Reports, 15 April 2011, p. 346) argues that the phonological complexity of languages reflects the loss of phonemic distinctions due to successive founder events during human migration (the serial founder hypothesis). Statistical simulations show that the type I error rate of Atkinson’s analysis is hugely inflated. The data at best support only a weak interpretation of the serial founder hypothesis.
Klein, N., Carlson, G., Li, R., Jaeger, T.F., and Tanenhaus, M. (in press) Classifying and Massifying Incrementally in Chinese Language Comprehension.
Submitted for proofs. Feel free to cite, but pls quote only from final copy.
Qian, T. and Jaeger, T. F. (in press). Cue Effectiveness in Communicatively Efficient Discourse Production. Cognitive Science.
Submitted for proofs. Feel free to cite, but pls quote only from final copy.
Recent years have seen a surge in accounts motivated by information theory that consider language production to be partially driven by a preference for communicative efficiency. Evidence from discourse production (i.e. production beyond the sentence level) has been argued to suggest that speakers distribute information across discourse so as to hold the conditional per-word entropy associated with each word constant, which would facilitate efficient information transfer (Genzel & Charniak, 2002). This hypothesis implies that the conditional (contextualized) probabilities of linguistic units affect speakers’ preferences during production. Here, we extend this work in two ways. First, we explore how preceding cues are integrated into contextualized probabilities, a question which so far has received little to no attention. Specifically, we investigate how a cue’s maximal informativity about upcoming words (the cue’s effectiveness) decays as a function of the cue’s recency. Based on properties of linguistic discourses as well as properties of human memory, we analytically derive a model of cue effectiveness decay and evaluate it against cross-linguistic data from twelve languages. Second, we relate the information theoretic accounts of discourse production to well-established mechanistic (activation-based) accounts: we relate contextualized probability distributions over words to their relative activation in a lexical network given
preceding discourse.
79 views
Seen by:
Jaeger, T.F., Furth, K., and Hilliard, C. (in press). Phonological overlap affects lexical selection during sentence production. Journal of Experimental Psychology: Learning, Memory, and Cognition.
Submitted for proofs. Feel free to cite, but pls quote only from final copy.
Theories of lexical production differ in whether they allow phonological processes to affect lexical selection directly. While some accounts, such as interactive activation accounts, predict (weak) early effects of phonological processes during lexical selection via feedback connections, strictly serial architectures do not make this prediction. We present evidence from lexical selection during unscripted sentence production that lexical selection is affected by the phonological form of recently produced words. In a video description experiment, participants described scenes that were compatible with several near meaning-equivalent verbs. We find that speakers are less likely than expected by chance to select a verb form that would result in phonological onset overlap with the subject of the sentence. Additional evidence from the distribution of disfluencies immediately preceding the verb argues that this effect is due to early effects on lexical selection, rather than later corrective processes, such as self-monitoring. Taken together, these findings support accounts that allow early feedback from phonological processes to word level nodes, even during lexical selection.
Butler, L. K., Jaeger, T. F., Bohnemeyer, J. and Furth, K. 2011. Learning to Express Visual Contrasts in the Production of Referring Expressions in Yucatec Maya. Proceedings of the PRE-CogSci 2011 workshop: Bridging the gap between computational, empirical and theoretical approaches to reference, Boston, MA, July 20th.
We examined rates of informativeness in the production of
modifications in response to a visual contrast In a video description task with speakers of Yucatec Maya. We analyzed modifications of referring expressions on the part of a speaker, and we also examined the effect of over- and under-informativeness on the listener’s comprehension. We found that prior experience with difficult comprehension did not significantly affect the listener’s rate of informativeness when in the role of speaker, but we found that experience as a speaker did result in reduced rates of under-informativeness. That is to say, as a speaker’s own experience progressed, the speaker became less under-informative. We discuss these results as audience design-based learning.
Keywords: referring expressions, informativeness, audience
design, learning, Yucatec Maya, field-based psycholinguistics
62 views
Seen by:
Croft,W., Bhattacharya, T., Kleinschmidt, D., Smith, D. E., and Jaeger, T. F. 2011. Greenbergian universals, diachrony and statistical analyses. Linguistic Typology 15(2), 433-453.
Submitted for proofs. Feel free to cite, but pls quote only from final copy.
Dunn et al. (2011) use a dynamic model of word order correlations in four phylogenies (language families) to argue that Greenbergian word order correlations are lineage-specific rather than universal. Dunn et al.’s model represents an important advance for diachronic typology (specifically, the dynamicization of a synchronic typology). However, certain assumptions made by Dunn and colleagues in the application of the model pose serious issues in accepting the conclusions, notably the absence of any Type II error analysis to assess the rate of false negatives, the absence of contact effects and the nature of the phylogenies used. Nevertheless, typologists should welcome the model and encourage the development of a revised model with more linguistically plausible assumptions.
Jaeger, T. F., Graff, P., Croft, B., and Pontillo, D. 2011. Mixed effect models for genetic and areal dependencies in linguistic typology: Commentary on Atkinson. Linguistic Typology 15(2), 281–319.
This is the author's copy. It is posted her for personal use, not redistribution. By downloading this manuscript you request an offprint from the author and agree to only use it for non-commercial use. This paper was published in inguistic Typology 15(2), 281–319.
Atkinson (2011) presents a large-scale quantitative test of the serial founder effect against cross-linguistic data on phonological diversity. This article has sparked considerable discussion. Here, we focus on potential issues with the statistical procedures employed in the paper. In particular, we investigate to what extent the results are robust once genealogical and geographic relations between languages is taken into account. To this end, we provide an introduction to linear mixed regression, the method employed by Atkinson to control for genealogical dependencies in the data. We illustrate why linear mixed modes are preferable to alternative proposals, such as separate by-group regressions (e.g. by language family or by continent). We then present two simple ways to incorporate controls for effects of language contact that, we hope, are of interest beyond the current commentary. Finally, we present simulations that assess the Type I error rate of Atkinson’s approach to find the most probable origin of language (i.e. false rejections of the null hypothesis).
Tily, H. and Jaeger, T.F. 2011. Complementing quantitative typology with behavioral approaches: Evidence for typological universals. Linguistic Typology 15(2), 497-508.
Submitted for proofs. Feel free to cite, but pls quote only from final copy.
Dunn et al. (2011) employ computational phylogenetic methods to test whether certain pairs of languages features are universally related in that they co-develop over time (Greenbergian implicational universals). They nd little evidence for universal word-order correlations, contrary to both generative and functional accounts of language. Other commentaries in this issue point to potential problems with the approach employed by Dunn and colleagues (e.g. Croft et al, this issue). Some of these are inherent to quantitative typology: in particular, sparsity of available data and uncertainty about language history. There are, however alternative methods for creating new data to test universal biases for certain word orders. Here we discuss two methods that we take to be of particular promise: Arti cial Language Learning, which has been used to study language acquisition, and Iterative Arti cial Language Learning, which extends the former method to the study of language change over generations. We discuss recent work within these two paradigms that suggests language learners exhibit universal biases that might cause universals like those discussed by Dunn et al. to emerge over time.
108 views
Seen by:
Kleinschmidt, D. & Jaeger, T.F. (2011). A Bayesian belief updating model of phonetic recalibration and selective adaptation. ACL Workshop on Cognitive Modeling and Computational Linguistics.
For page numbers, please see the final version on the CMCL workshop at the 2011 ACL: http://www.psy.cmu.edu/~cmcl/
The mapping from phonetic categories to acoustic cue values is highly flexible, and adapts rapidly in response to exposure. There is currently, however, no theoretical framework which captures the range of this adaptation. We develop a novel approach to modeling phonetic adaptation via a belief-updating model, and demonstrate that this model naturally unifies two adaptation phenomena traditionally considered to be distinct.
Tily, H.J., Frank, M.C. & Jaeger, T.F. (2011). The learnability of constructed languages reflects typological patterns. The 33rd Annual Meeting of the Cognitive Science Society (CogSci11). Boston, MA. July, 2011, 1364-1369.
A small number of the logically possible word order configurations
account for a large proportion of actual human languages.
To explain this distribution, typologists often invoke principles
of human cognition which might make certain orders easier or
harder to learn or use. We present a novel method for carrying
out very large scale artificial language learning tasks over
the internet, which allows us to test large batteries of systematically designed languages for differential learnability. An exploratory study of the learnability of all possible configurations
of subject, verb, and object finds that the two most frequent
orders in human languages are the most easily learned, and
yields suggestive evidence compatible with other typological
and psycholinguistic observations.
Keywords: artificial grammar; language acquisition; language
typology; psycholinguistics; word order
121 views
Seen by:
Hilliard, C., Furth, K. & Jaeger, T.F. (2011). Phonological Encoding in Sentence Production. The 33rd Annual Meeting of the Cognitive Science Society (CogSci11). Boston, MA. July, 2011, 3070-3075.
Previous tests of the phonological competition model (Dell, 1986) have mostly investigated the effects of phonological overlap (e.g. pick-pin) in isolated word production (e.g. primed picture naming). This is problematic since recent findings suggest that the effect of phonological overlap depends on the syntactic category of the phonologically related words, and few previous studies investigate phonological planning in the context of grammatical strings. We introduce a novel paradigm to examine two predictions of the so called parallel-then-sequential competition model (O‟Seaghdha and Marin, 2000) against data from the distribution of disfluencies in sentence production. We also extend previous work by comparing different forms of phonological overlap (identity vs. similarity) in both word onsets and rhymes.
Keywords: phonological encoding; sentence production
Fine, A.B. & Jaeger, T.F. (2011). Language comprehension is sensitive to changes in the reliability of lexical cues. The 33rd Annual Meeting of the Cognitive Science Society (CogSci11). Boston, MA. July, 2011, 925-930.
This paper tests the hypothesis that language comprehenders
update their beliefs about the statistics of their language
throughout the lifespan, and that this belief update allows comprehenders to combine probabilistic linguistic cues according to their reliability. We conduct a multi-day sentence comprehension study in which the reliability of a probabilistic cue to syntactic structure is manipulated between subjects. We\ find that as the reliability of one cue to syntactic structure decreases, comprehenders come to rely more on a second cue to syntactic structure. The results are consonant with rational models of cue integration in speech perception and in nonlinguistic domains, thus suggesting a unifying computational principle governing the way humans use information across both perceptual and higher-level cognitive tasks.
Keywords: psycholinguistics; adaptation; sentence processing;
cue combination
Fedzechkina, M., Jaeger, T.F. & Newport, E. (2011). Functional Biases in Language Learning: Evidence from Word Order and Case-Marking Interaction. The 33rd Annual Meeting of the Cognitive Science Society (CogSci11). Boston, MA. July, 2011, 318-323.
Why do languages share structural properties? The functionalist tradition has argued that languages have evolved to suit the needs of their users. By what means functional pressures may come to shape grammar over time, however, remains unknown. Functional pressures could affect adults’ production; or they could operate during language learning. To date, these possibilities have remained largely untested. We explore the latter possibility, that functional pressures operate during language acquisition. In an artificial language learning experiment we investigate the trade-off between word order and case. Flexible word order languages are potentially ambiguous if no case-marking (or other cues) are employed to identify the doer of the action. We explore whether language learners are biased against uncertainty in the mapping of form and meaning, showing a tendency to make word order a stronger cue to the intended meaning in no-case languages.
Keywords: Language acquisition; language universals; acquisition
biases; word order; case-marking
141 views
Seen by:
Farmer, T., Fine, A.B. & Jaeger, T.F. (2011). Implicit Context-Specific Learning Leads to Rapid Shifts in Syntactic Expectations. The 33rd Annual Meeting of the Cognitive Science Society (CogSci11). Boston, MA. July, 2011, 2055-2060..
During incremental language understanding, comprehenders draw on a rich base of probabilistic cues to efficiently process the noisy perceptual input they receive. One challenge listeners face in employing such cues is that most cues are context-dependent. Here, we present an experiment that investigates the extent to which listeners learn situation-specific adjustments in the information and/or weight of the lexical bias of a verb. Specifically, we ask to what extent comprehenders are able to rapidly change their interpretation of lexical cues to syntactic structure, where such behavior would be rational due to situation-specific statistics in the environment.
Keywords: Language Comprehension; Ambiguity Resolution;
Learning Effects; Language Experience
Jaeger, T. F. and Tily, H. 2011. Language Processing Complexity and Communicative Efficiency. WIRE: Cognitive Science 2(3), 323-335.
This is a pre-final draft. For the final version, please visit http://onlinelibrary.wiley.com/doi/10.1002/wcs.126/full
Functionalist typologists have long argued that pressures associated with language usage influence the distribution of grammatical properties across the world’s languages. Specifically, grammatical properties may be observed more often across languages because they improve a language’s utility or decrease its complexity. While this approach to the study of typology offers the potential of explaining grammatical patterns in terms of general principles rather than domain-specific constraints, the notions of utility and complexity are more often grounded in intuition than empirical findings. A suitable empirical foundation might be found in the terms of processing preferences: in that case, psycholinguistic measures of complexity are then expected correlate with typological patterns. We summarize half a century of psycholinguistic work on ‘processing complexity’ in an attempt to make this work accessible to a broader audience:
What makes something hard to process for comprehenders, and what determines speakers’ preferences in production? We also briefly discuss recently emerging approaches that link preferences in production to communicative efficiency. These
approaches can be seen as providing well-defined measures of utility. With these psycholinguistic findings in mind, it is possible to investigate the extent to which language usage is reflected in typological patterns. We close with a summary of paradigms that allow the link between language usage and typology to be studied empirically.
Hofmeister, P., Jaeger, T. F., Arnon, I., Sag, I.A., and Snider, N. in press. The source ambiguity problem: distinguishing the effects of grammar and processing on acceptability judgments. Language and Cognitive Processes.
I've updated this file 07/05/2011. This is a pre-final draft. For off-prints, please email me or go to the LCP page.
Judgments of linguistic unacceptability may theoretically arise from either grammatical deviance or significant processing difficulty. Acceptability data are thus naturally ambiguous in theories that explicitly distinguish formal and functional constraints. Here, we consider this source ambiguity problem in the context of Superiority effects: the dispreference for ordering a wh-phrase in front of a syntactically “superior” wh-phrase in multiple wh-questions, e.g. What did who buy? More specifically, we consider the acceptability contrast between such examples and so-called D-linked examples, e.g. Which toys did which parents buy? Evidence from acceptability and self-paced reading experiments demonstrates that (i) judgments and processing times for Superiority violations vary in parallel, as determined by the kind of wh-phrases they contain, (ii) judgments increase with exposure while processing times decrease, (iii) reading times are highly predictive of acceptability judgments for the same items, and (iv) the effects of the complexity of the wh-phrases combine in both acceptability judgments and reading times. This evidence supports the conclusion that D-linking effects are likely reducible to independently motivated cognitive mechanisms whose effects emerge in a wide range of sentence contexts. This in turn suggests that Superiority effects, in general, may owe their character to differential processing difficulty
Wasow, T., Jaeger, T. F., and Orr, D. 2011. Lexical Variation in Relativizer Frequency. In H. Simon and H. Wiese (eds.): Proceedings of the 2005 DGfS workshop “Expecting the unexpected: Exceptions in Grammar”, 175-196. Berlin/New York: De Gruyter Mouton
An exception to a non-categorical generalization consists of a lexical item that exhibits the general pattern at a rate radically different – either far higher or far lower – from the norm. Lexical differences in noun phrases containing non-subject relative clauses (NSRCs) correlate with large differences in the likelihood that the NSRC will begin with that. In particular, the choices of determiner, head noun, and prenominal adjective in an NP containing an NSRC may dramatically raise or lower rates of that in the NSRC. These lexical variations can be partially explained in terms of predictability: more predictable NSRCs are less likely to begin with that. This generalization can be plausibly explained in terms of processing, assuming that facilitates processing and/or signals difficulty. The correlations between lexical choices in the NP and the predictability of an NSRC can, in turn, be explained in terms of the semantics of the lexical items and the pragmatics of reference.
Fine, A. B., Qian, T., Jaeger, T. F., and Jacobs, R. 2010. Is there syntactic adaptation in language comprehension. Proceedings of ACL: Workshop on Cognitive Modeling and Computational Linguistics, 18-26.Uppsala, Sweden.
We're building on this paper in a recent (2011) CogSci submission: Farmer, Fine, and Jaeger, which is available on this page.
In this paper we investigate the manner in which the human language comprehension system adapts to shifts in probability distributions over syntactic structures, given experimentally controlled experience with those structures. We replicate a classic reading experiment, and present a model of the behavioral data that implements a form of Bayesian belief update over the course of the experiment.
70 views
Seen by:
Jaeger, T. F. 2010. Redundancy and Reduction: Speakers Manage Syntactic Information Density. Cognitive Psychology, 61(1), 23-62.
Downloading this file constitutes a request for an off-print from the author. Thank you for your interest in my work.
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.
Cook, S. W., Jaeger, T. F., and Tanenhaus, M. 2009. Producing Less Preferred Structures: More Gestures, Less Fluency. The 31st Annual Meeting of the Cognitive Science Society (CogSci09), 62-67.
http://www.hlp.rochester.edu/publications/CookJaegerTanenhaus_cogsci09
Speakers often have choices about how to structure their utterances. However, even though multiple alternatives may be acceptable in theory, often one of them will be preferred over the others. The question we explored here was what happens when speakers produce less preferred alternatives. We developed a new experimental paradigm to reliably elicit the propositional or double object dative with varying degrees of preference. We then used this paradigm to investigate how, given properties of the message, an individual speaker's preference for a particular structure affects how that utterance is produced. Speakers gestured more and were more likely to be disfluent when they chose less preferred structures. Thus, having a choice per se does not guarantee more successful production. Instead, production is facilitated when speakers choose more preferred alternatives.
25 views
Seen by:
Qian, T. and Jaeger, T. F. 2009. Evidence for Efficient Language Production in Chinese. The 31st Annual Meeting of the Cognitive Science Society (CogSci09), 851-856.
http://www.hlp.rochester.edu/publications/Qian_Jaeger_cogsci09.pdf
Recent work proposes that language production is organized to facilitate efficient communication by means of transmitting information at a constant rate. However, evidence has almost exclusively come from English. We present new results from Mandarin Chinese supporting the hypothesis that Constant Entropy Rate is observed cross-linguistically, and may be a universal property of the language production system. We show that this result holds even if several important confounds that previous work failed to address are controlled for. Finally, we present evidence that Constant Entropy Rate is observed at the syllable level as well as the word level, suggesting findings do not depend on the chosen unit of observation.
8 views
Seen by: Jaeger, T. F. 2008. Categorical Data Analysis: Away from ANOVAs (transformation or not) and towards Logit Mixed Models. Journal of Memory and Language 59, 434-446.
Downloading this file constitutes a request for an off-print from the author. Thank you for your interest in my work.
The code and data mentioned in this paper can be found at this HLP/Jaeger Lab blog post: http://hlplab.wordpress.com/2009/04/20/jaeger08/
The post also contains links to further tutorials and reading materials.
This paper identifies several serious problems with the widespread use of ANOVAs for the analysis of categorical outcome variables such as forced-choice variables, question-answer accuracy, choice in production (e.g. in syntactic priming research), et cetera. I show that even after applying the arcsine-square-root transformation to proportional data, ANOVA can yield spurious results. I discuss conceptual issues underlying these problems and alternatives provided by modern statistics. Specifically, I introduce ordinary logit models (i.e. logistic regression), which are well-suited to analyze categorical data and offer many advantages over ANOVA. Unfortunately, ordinary logit models do not include random effect modeling. To address this issue, I describe mixed logit models (Generalized Linear Mixed Models for binomially distributed outcomes, Breslow and Clayton [Breslow, N. E. & Clayton, D. G. (1993). Approximate inference in generalized linear mixed models. Journal of the American Statistical Society 88(421), 9–25]), which combine the advantages of ordinary logit models with the ability to account for random subject and item effects in one step of analysis. Throughout the paper, I use a psycholinguistic data set to compare the different statistical methods.
Jaeger, T. F. and Snider, N. 2008. Implicit learning and syntactic persistence: Surprisal and Cumulativity. The 30th Annual Meeting of the Cognitive Science Society (CogSci08), 1061-1066.
http://www.bcs.rochester.edu/people/fjaeger/papers/E_JaegerSnider08cog
Speakers often repeat syntactic structures that have recently been used by them or their interlocutors. Such syntactic persistence has variously been attributed to social causes (mimicking to be well-liked), to an effort to reduce processing costs, or as a side effect of either transient activation of previously processed structure or implicit learning. We present new evidence from syntactic persistence in conversational speech consistent with the view that syntactic persistence is an epiphenomenon. We propose that syntactic persistence is linked to the updating and maintenance of probabilistic syntactic knowledge. We find that syntactic persistence is SURPRISAL-SENSITIVE and CUMULATIVE: Syntactic persistence is stronger the less probable the prime and it is sufficiently long-lived to be cumulative.
Walter, M.A. and Jaeger, T. F. 2008. Constraints on English that-drop: A strong lexical OCP effect. In Edwards, R.L., Midtlyng, P.J., Stensrud, K.G., and Sprague, C.L. (eds.) Proceedings of the Main Session of the 41st Meeting of the Chicago Linguistic Society, 505-519. Chicago, IL: CLS.
This is a pre-final draft. The final draft will appear (in some far point in the future) in the CLS proceedings.
This paper investigates speakers' dispreference to produce two identical word forms in adjacency and how this dispreference affects that-mentioning in complement and relative clauses. See also Jaeger, T. F. in press. Phonological Optimization and Syntactic Variation: The Case of Optional that. Proceedings of the 32nd Annual Meeting of the Berkeley Linguistic Society.
Frank, A. and Jaeger, T. F. 2008. Speaking Rationally: Uniform Information Density as an Optimal Strategy for Language Production. The 30th Annual Meeting of the Cognitive Science Society (CogSci08), 939-944.
http://www.bcs.rochester.edu/people/fjaeger/papers/FrankJaeger08cogsci
We provide evidence for a rational account of language production, Uniform Information Density (UID, Jaeger, 2006; Levy & Jaeger, 2007). Under the assumption that communication can usefully be understood as information transmission over a capacity-limited noisy channel, an optimal strategy in language production is to maintain a uniform rate of information transmission close to the channel capacity. This theory predicts that speakers will make strategic use of the flexibility allowed by their languages. Speakers should plan their utterances so that elements with high information are lengthened, and elements with low information are shortened, making the amount of information transmitted per time more uniform (and hence closer to the optimum). In three corpus studies, we show that American English speakers’ use of contractions (“you are”! “you’re”) follows the predictions of UID. We then explore further implications of UID for production planning. Keywords: language production; utterance planning; information theory; morphological reduction; contractions
Gómez Gallo, C., Jaeger, T. F., and Smyth, R. 2008. Incremental Syntactic Planning across Clauses. The 30th Annual Meeting of the Cognitive Science Society (CogSci08), 1294-1299.
This is a pref-final draft. The final draft is available through the CogSci online proceedings.
Language production research has focused on planning at the lexical and syntactic levels. Consequently, little is known about speakers’ strategies for distributing information across multiple utterances. We present evidence that (a) the overall amount of information in an intended message affects how speakers distribute that information across clauses, and (b) speakers have relatively early access to (at least an estimate of) message complexity, specifically before word retrieval. The observed effect is unexpected for competing theories of sentence production based on availability or macro-propositional accounts. We discuss these findings in an information theoretic context coupled with a sketch of a model of limited resources at the level of macro-propositional planning.
35 views
Seen by: Levy, R. and Jaeger, T. F. 2007. Speakers optimize information density through syntactic reduction. In B. Schlökopf, J. Platt, and T. Hoffman (Eds.), Advances in neural information processing systems (NIPS) 19, 849-856. Cambridge, MA: MIT Press.
http://www.bcs.rochester.edu/people/fjaeger/papers/E_LevyJaeger07_full
If language users are rational, they might choose to structure their utterances so as to optimize communicative properties. In particular, information-theoretic and psycholinguistic considerations suggest that this may includemaximizing the uniformity of information density in an utterance. We investigate this possibility in the context of syntactic reduction, where the speaker has the option of either marking a higher-order unit (a phrase) with an extra word, or leaving it unmarked. We demonstrate that speakers are more likely to reduce less information-dense phrases. In a second step, we combine a stochastic model of structured utterance production with a logistic-regression model of syntactic reduction to study which types of cues speakers employ when estimating the predictability of upcoming elements. We demonstrate that the trend toward predictability-sensitive syntactic reduction (Jaeger, 2006) is robust in the face of a wide variety of control variables, and present evidence that speakers use both surface and structural cues for
predictability estimation.
62 views
Seen by:
Cross-linguistic Variation in a Processing Account: The Case of Multiple Wh-questions
published in BLS proceedings 2006, co-authored with Inbal Arnon, Neal Snider, Philip Hofmeister, T. Florian Jaeger & Ivan A. Sag
23 views
Seen by: Accent-free prosodic phrases? Accents and phrasing in the post-nuclear domain
http://www.bcs.rochester.edu/people/fjaeger/papers/is2005_norcliffejae
Some theories of English intonation, e.g. [1], [2], posit a crucial connection between pitch accent placement and the identification of prosodic structure, stating that every prosodic phrase must contain at least one pitch accent. We provide new evidence from the empirical study of prosodic phrasing in the post-nuclear domain to show that this view cannot be maintained. Our experimental study, based on the prosodic disambiguation of verb-particle vs. verb-preposition readings in the post-nuclear domain, revealed significant length based prosodic differences between the two readings. This argues that the post nuclear domain is divided into more than one prosodic phrase, the second of which doesn’t bear any pitch accent.
Optional that indicates production difficulty: Evidence from disfluencies
http://www.bcs.rochester.edu/people/fjaeger/papers/E_Jaeger_2005_DiSS0
Optional word omission, such as that omission in complement and relative clauses, has been argued to be driven by production pressure (rather than by comprehension). One particularly strong production-driven hypothesis states that speakers insert words to buy time to alleviate production difficulties. I present evidence from the distribution of disfluencies in non-subject extracted relative clauses arguing against this hypothesis. While word omission is driven by production difficulties, speakers may use that as a collateral signal to addressees, informing them of anticipated production difficulties. In that sense, word omission would be subject to audience design (i.e.catering to addressees’ needs).
45 views
Seen by:
Topics First! In- and outside of Bulgarian wh-interrogatives
http://www.bcs.rochester.edu/people/fjaeger/papers/E_Jaeger03_HPSG.pdf
In Jaeger (to appear) I have described clitic doubling in Bulgarian whinterrogatives which constitutes a type of Superiority violation that cannot be accounted for by any existing analyses. By showing that clitic doubling of object wh-phrases marks topicality, I raised the hypothesis that many (or maybe all) so called Superiority effects in Bulgarian are due to topic-fronting of wh-phrases. Here, I provide further support for this hypothesis and show that there is also evidence for topic-fronting of non-object wh-phrases. Differences between colloquial and formal Bulgarian are restricted to how topical objects have to be realized at the site of the extraction (i.e. the VP), which also makes the account readily extendable to other multiple fronting languages. The complex ordering constraints on the left periphery are captured in a Linear Syntax approach (similar to but different from Kathol 2000).
38 views
Seen by:
Towards a Dynamic Model of Topic-Marking
http://www.bcs.rochester.edu/people/fjaeger/papers/Topic_final_preproc
This paper addresses the gap between the research on D[iscourse] S[tructure] T[opics] (cf. Chafe 2001, Fraser 1988) and I[nformation] S[tructure] T[opics] (cf. Jacobs 2001, Büring 1999, Lambrecht 1994, Vallduvi 1992). This paper takes a uniform approach to topics where ISTs and DSTs are treated as ontologically identical. We propose a hierarchical model of T(opic)-trees, which are constructed dynamically in the course of discourse. The nodes of a T-tree consist of sets of semantic objects (e.g. entities, propositions, predicates, etc.; contrasting with Dik's 1989 D-topics). The higher a T-tree node is in the tree, the bigger is the linguistic form unit that it corresponds to (e.g. the whole text, a paragraph, a sentence). The highest topic in a T-tree is always the global DST, the lowest topics are the ISTs. Based on T-trees, we show how the (topic-)coherence of texts could be modelled. The T-trees are motivated by two case studies, one on the 'as for' construction in English and one on multiple topic constructions in Japanese. The proposed model allows us to formalize the constraints which those constructions put on the context in which they appear. The remainder of the paper, discusses the interaction of the proposed model with the notion of focus and kontrast (cf. Vallduví and Vilkuna 1998) making reference to such various constructions as left-dislocation, itclefts, and topicalization in English, as well as, Multiple Nominative Constructions in Japanese.
BULGARIAN WORD ORDER AND THE ROLE OF THE DIRECT OBJECT CLITIC IN LFG
http://cslipublications.stanford.edu/LFG/7/lfg02jaegergerassimova.pdf
This paper provides an LFG account of the Bulgarian direct object clitic's interaction with information structure (i.e. topic-focus structure) and word order. We show that the direct object clitic has at least two functions (it is both a topical object agreement marker and default pronoun) and then demonstrate how our account correctly predicts in which syntactic environment which of the two functions can be chosen. In order to achieve this we allow for two different ways to identify a 'topic' in LFG – a move, which reduces the necessary claims about the direct object clitic's behaviour to the most general principles of LFG (i.e. Uniqueness, Completeness, Extended Coherence). The proposed analysis is based on extensive evidence (our own online experiment, Leafgren 1997a,b, 1998, and Avgustinova 1997), and incorporates recent findings on the discourse-configurationality of the left periphery in Bulgarian clauses (cf. Rudin 1997, Arnaudova 2001, Dimitrova-Vulchanova & Hellan 1998). Although covering a much broader range of data from spoken Bulgarian than other formal accounts, our account makes the right predictions about possible word orders and the optional, or obligatory presence/absence of the direct object clitic. Unlike almost all other recent accounts, our analysis does not rely on the assumption of configurationality, which has been shown to be problematic for Bulgarian (cf. Gerassimova & Jaeger 2002).
Configurationality and the Direct Object Clitic in Bulgarian
http://www.bcs.rochester.edu/people/fjaeger/papers/E_GerassJaeger02_ES
Bulgarian, as a free word order language without case, presents an interesting problem for the ongoing discussion whether or to what extent all languages are configurational (as defined in
Nordlinger 1998:45). In this paper, we discuss the typological status of Bulgarian regarding its configurationality. First, we apply a set of configurationality tests (cf. King 1995, Speas 1990). Second, we show how the direct object clitic figures in determining possible constituent orders. Related to the direct object clitic, the role of information structure proves to be of crucial importance for the determination of word order in Bulgarian.
38 views
Seen by: