Linguistics Disfluency
by
Scott H. Fraundorf, Jennifer Arnold
  • LAST MODIFIED: 28 April 2014
  • DOI: 10.1093/obo/9780199772810-0189

Introduction

Disfluencies are interruptions in the regular flow of speech, such as using uh and um, pausing silently, repeating words, or interrupting oneself to correct something said previously. Disfluency can be distinguished from speech errors in which the speaker produces the wrong words or speech sounds but may doE so without any interruptions in the flow of speech. Disfluencies commonly stem from delays or errors in the cognitive processes of language production, although some studies indicate that social or other cognitive factors may play a role too. Even in non-disordered speech, disfluency is common and thus relevant to general theories of language production and comprehension. However, the frequency and types of disfluency may also be influenced by various clinical diagnoses (sometimes separately termed “dysfluency”); these include most notably stuttering, but also attention deficit / hyperactive disorder and autism spectrum disorders. Children’s speech also includes disfluencies, but disfluency does not necessarily indicate poor language acquisition; in fact, some disfluencies are used more by children who are at a more advanced stage of linguistic or conceptual development. Some work has considered whether certain disfluencies—namely, fillers such as uh and um—should be considered deliberate signals produced by a speaker to indicate having momentary trouble with language production, but this hypothesis remains debated. Although disfluencies interrupt the flow of speech, they do not necessarily impair comprehension; in fact, many studies have observed that disfluency can even facilitate comprehension in some situations, because disfluencies increase attention to the speech stream or because they allow a listener to predict that the speaker will next refer to something difficult. However, when speakers correct themselves, the original, erroneous material is not always completely forgotten by hearers, and there are lingering effects in comprehension of the original material. Moreover, disfluency can present a challenge both for automated speech-recognition systems and text-to-speech systems, and many computational studies have examined models for detecting and accommodating disfluency.

General Overviews

Several articles have reviewed the disfluency literature, with varying aims. Ferreira and Bailey 2004 reviews the comprehension of disfluent speech and its consequences for syntactic processing. Corley and Stewart 2008 reviews the literature on filled pauses in particular, and the authors use the literature to argue against the claim that filled pauses constitute a deliberate linguistic signal.

  • Corley, M., and O. W. Stewart. 2008. Hesitation disfluencies in spontaneous speech: The meaning of um. Language and Linguistics Compass 2.4: 589–602.

    DOI: 10.1111/j.1749-818X.2008.00068.xSave Citation »Export Citation »E-mail Citation »

    Clear review of production and comprehension of filled pauses such as um. On the basis of this evidence, presents the argument that filled pauses do not constitute deliberately planned linguistic signals.

    Find this resource:

    • Ferreira, F., and K. G. D. Bailey. 2004. Disfluencies and human language comprehension. Trends in Cognitive Sciences 8.5: 231–237.

      DOI: 10.1016/j.tics.2004.03.011Save Citation »Export Citation »E-mail Citation »

      Provides an accessible introduction to disfluencies and a short review of the comprehension of disfluent speech. Has a particular focus on syntactic processing of disfluent speech.

      Find this resource:

      Books

      Several books addressing broad theories of language behavior devote attention to disfluency. Levelt 1993 examines disfluency in the context of the cognitive process of language production, while Clark 1996 considers how disfluency is used in coordinating the interactive process of conversation. Meanwhile, Erard 2007 presents a popular, nontechnical overview of research on disfluency.

      • Clark, H. H. 1996. Using language. Cambridge, UK: Cambridge Univ. Press.

        DOI: 10.1017/CBO9780511620539Save Citation »Export Citation »E-mail Citation »

        Presents Clark’s view of language as joint action between multiple interlocutors. Considers how delays in production present problems both for speakers and hearers and how various disfluencies represent actions on a secondary conversational track to manage the issue. Important for considering disfluencies in the primary context of language use: conversation.

        Find this resource:

        • Erard, M. 2007. Um . . .: Slips, stumbles, and verbal blunders, and what they mean. New York: Pantheon.

          Save Citation »Export Citation »E-mail Citation »

          Popular press book reviewing key findings regarding both disfluencies and speech errors. Provides an accessible and accurate introduction to disfluencies, although no primary data or theoretical innovations.

          Find this resource:

          • Levelt, W. J. M. 1993. Speaking: From intention to articulation. ACL-MIT Press Series in Natural-Language Processing. Cambridge, MA: MIT Press.

            Save Citation »Export Citation »E-mail Citation »

            Outlines Levelt’s comprehensive theory of the cognitive processes of language production. Considers how disfluency can result from deviations from accurate production, and attempts to monitor or correct those problems. Important because of its consideration of disfluency itself and because theories of fluent language production provide a foundation for studying disfluency. First published in 1989; reprinted as recently as 2007.

            Find this resource:

            • Segalowitz, N. 2010. Cognitive bases of second language fluency. Cognitive Sciences and Second Language Acquisition. New York: Routledge.

              Save Citation »Export Citation »E-mail Citation »

              Discusses the factors that underlie fluency (or lack thereof) in a second language and how disfluency can inform second-language research and assessment.

              Find this resource:

              Journals

              No journal specializes in disfluency in normal language production, but disfluency is a frequent topic of investigation in journals related to clinical aspects of language production, such as the Journal of Fluency Disorders.

              In Language Comprehension

              Several studies examine how disfluency affects language comprehension. These questions are important for understanding the architecture of the language-processing system in humans, and also for the purpose of building computational models that can deal with disfluent input. Papers are categorized around four questions: (1) Can listeners detect disfluency? (2) How does disfluency affect speech comprehension? (3) How does disfluency serve as a cue that the speaker is correcting an error? (4) How do listeners use disfluency to infer the speaker’s metacognitive state?

              Detecting Disfluencies in the Speech Input

              Lickley and Bard 1998 addresses one of the most basic questions about how disfluency affects language comprehension: whether listeners recognize disfluent input when they hear it.

              • Lickley, R. J., and E. G. Bard. 1998. When can listeners detect disfluency in spontaneous speech? Language and Speech 41.2: 203–226.

                DOI: 10.1177/002383099804100204Save Citation »Export Citation »E-mail Citation »

                Three experiments test whether human comprehenders can detect disfluency in speech input. Finds that listeners were not adept at predicting disfluencies, but they did reliably notice disfluencies once they had occurred, sometimes even before identifying the word.

                Find this resource:

                Effects on Speech Comprehension

                A major focus of recent research in psycholinguistics concerns the effect that disfluency has on other aspects of speech comprehension. Does disfluency hinder effective comprehension? Or does it actually help, and by what cognitive processes? These papers are divided into sections on the four following themes: (1) word recognition, (2) attention and memory, (3) disfluency as a cue, and (4) parsing.

                Word Recognition

                Studies in this section examine whether the presence of disfluency affects the listener’s ability to recognize words in the speech stream. While Fox Tree 1995 reports inhibitory effects of disfluency on word identification, Corley and Hartsuiker 2011 reports facilitory effects from temporal delays, including those due to disfluency.

                Attention and Memory

                A number of papers present experimental evidence in support of the hypothesis that disfluency affects language comprehension by influencing listeners’ attention to the speech input, and resulting memory representations. Fox Tree 2001 uses a word identification task to examine how the presence of ums and uhs affects word recognition, and the author interprets her results in terms of the effect of disfluency on attention to the speech input. Collard, et al. 2008 extends Fox Tree’s work by using event-related potentials (ERPs) to show that filled pauses elicit classic attentional components. Studies in Corley, et al. 2007; MacGregor, et al. 2009; and MacGregor, et al. 2010 also use ERPs to demonstrate that filled pauses and silent pauses modulate the class N400 effect, but disfluent repetitions do not. Fraundorf and Watson 2011 uses a memory task to demonstrate the effects of filled pauses on memory, and the authors also interpret their results in terms of the focusing effect that disfluency has on upcoming speech input.

                • Collard, P., M. Corley, L. J. MacGregor, and D. I. Donaldson. 2008. Attention orienting effects of hesitations in speech: Evidence from ERPs. Journal of Experimental Psychology: Learning, Memory, and Cognition 34.3: 696–702.

                  DOI: 10.1037/0278-7393.34.3.696Save Citation »Export Citation »E-mail Citation »

                  Uses ERPs to examine the suggestion in Fox Tree 2001 that filled pauses affect listeners’ attention to speech. Shows that classic attentional components (mismatch negativity and P300) are affected by disfluency, supporting an attentional-mechanism account of disfluency effects on comprehension.

                  Find this resource:

                  • Corley, M., L. J. MacGregor, and D. I. Donaldson. 2007. It’s the way that you, er, say it: Hesitations in speech affect language comprehension. Cognition 105.3: 658–668.

                    DOI: 10.1016/j.cognition.2006.10.010Save Citation »Export Citation »E-mail Citation »

                    Examines the classic ERP N400 effect, which is typically greater for unpredictable than predictable words. Shows that this effect is reduced if a disfluency precedes the critical word, and memory for the word is also improved. (See also As a Cue to the Speaker’s Meaning: Adults and As a Cue to the Speaker’s Meaning: Children.)

                    Find this resource:

                    • Fox Tree, J. E. 2001. Listeners’ uses of um and uh in speech comprehension. Memory & Cognition 29.2: 320–326.

                      DOI: 10.3758/BF03194926Save Citation »Export Citation »E-mail Citation »

                      This experimental study shows that the filled pause uh facilitates speech comprehension, whereas um has neither a facilitory nor an inhibitory effect. Fox Tree suggests that uh, which typically signals a short delay, may focus listeners’ attention on upcoming speech.

                      Find this resource:

                      • Fraundorf, S. H., and D. G. Watson. 2011. The disfluent discourse: Effects of filled pauses on recall. Journal of Memory and Language 65.2: 161–175.

                        DOI: 10.1016/j.jml.2011.03.004Save Citation »Export Citation »E-mail Citation »

                        Participants heard and retold stories that were manipulated to contain disfluent fillers (uh, um) or coughs. In two experiments, only disfluent fillers, and not non-speech disruptions, facilitated recall. It did not matter whether fillers were predictive of upcoming material. It is concluded that disfluency modulates attentional orienting to speech.

                        Find this resource:

                        • MacGregor, L. J., M. Corley, and D. I. Donaldson. 2009. Not all disfluencies are equal: The effects of disfluent repetitions on language comprehension. Brain and Language 111.1: 36–45.

                          DOI: 10.1016/j.bandl.2009.07.003Save Citation »Export Citation »E-mail Citation »

                          In contrast with Corley, et al. 2007 and MacGregor, et al. 2010, demonstrates that disfluent repetitions don’t modulate the N400 effect in the same way as filled pauses and silent pauses. However, disfluent repetitions did lead to a relative positivity at a later time window.

                          Find this resource:

                          • MacGregor, L. J., M. Corley, and D. I. Donaldson. 2010. Listening to the sound of silence: Disfluent silent pauses in speech have consequences for listeners. Neuropsychologia 48.14: 3982–3992.

                            DOI: 10.1016/j.neuropsychologia.2010.09.024Save Citation »Export Citation »E-mail Citation »

                            Extends the finding from Corley, et al. 2007, by showing that the N400 effect is also attenuated following silent pauses.

                            Find this resource:

                            As a Cue to the Speaker’s Meaning: Adults

                            A number of studies have examined whether disfluency has a specific effect on language comprehension by indicating to the listener something about the speaker’s meaning. Several studies demonstrate that disfluency leads listeners to expect a certain type of reference. For example, Arnold, et al. 2003 and Arnold, et al. 2004 demonstrate that adults have a disfluency bias toward discourse-new information; Arnold, et al. 2007 demonstrates that adults have a disfluency bias toward something conceptually or lexically novel. Barr and Seyfeddinipur 2010 shows that the disfluency bias to discourse-new information is driven by what is known to the speaker, and Arnold, et al. 2007 shows similar evidence for the disfluency bias to novel referents. Watanabe, et al. 2008 demonstrates a disfluency bias toward something structurally or conceptually complex, and Corley, et al. 2007 demonstrates a bias toward something unpredictable. Arnold and Tanenhaus 2011 and Barr 2001 link these findings to evidence that speakers are more likely to be disfluent in certain situations. Note that most of the work in this section does not assume that disfluency has to be an intentional signal from the speaker to be informative to listeners.

                            • Arnold, J. E., M. Fagnano, and M. K. Tanenhaus. 2003. Disfluencies signal theee, um, new information. Journal of Psycholinguistic Research 32.1: 25–36.

                              DOI: 10.1023/A:1021980931292Save Citation »Export Citation »E-mail Citation »

                              This paper reports preliminary evidence from the experiment in Arnold, et al. 2004 and discusses its findings in terms of the expectancy of referents.

                              Find this resource:

                              • Arnold, J. E., C. L. Hudson Kam, and M. K. Tanenhaus. 2007. If you say thee uh you are describing something hard: The on-line attribution of disfluency during reference comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition 33.5: 914–930.

                                DOI: 10.1037/0278-7393.33.5.914Save Citation »Export Citation »E-mail Citation »

                                Shows that disfluent utterances lead listeners to expect reference to something unfamiliar (e.g., a novel shape) rather than familiar (e.g., ice cream). The effect was reduced when listeners thought the speaker had object agnosia and perceived familiar objects as difficult, but not when listeners thought the speaker was distracted. Suggests that disfluency’s effects on comprehension involve at least some degree of attribution.

                                Find this resource:

                                • Arnold, J. E., and M. K. Tanenhaus. 2011. Disfluency effects in comprehension: How new information can become accessible. In The processing and acquisition of reference. Edited by E. Gibson and N. J. Pearlmutter, 197–217. Cambridge, MA: MIT Press.

                                  DOI: 10.7551/mitpress/9780262015127.001.0001Save Citation »Export Citation »E-mail Citation »

                                  Discusses possible mechanisms for the biases toward discourse-new and novel referents. Also includes corpus data to support the correlation between disfluency and reference to new information.

                                  Find this resource:

                                  • Arnold, J. E., M. K. Tanenhaus, R. J. Altmann, and M. Fagnano. 2004. The old and thee, uh, new: Disfluency and reference resolution. Psychological Science 15.9: 578–582.

                                    DOI: 10.1111/j.0956-7976.2004.00723.xSave Citation »Export Citation »E-mail Citation »

                                    Uses a visual-world eye-tracking paradigm. Listeners followed instructions such as “Put the grapes below the candle. Now put . . . thee . . . uh . . . {camel/candle} . . . .” Listeners looked to the new camel more quickly following disfluent speech, but to the given candle more quickly following fluent speech. Shows that disfluency leads comprehenders to expect reference to discourse-new information.

                                    Find this resource:

                                    • Barr, D. J. 2001. Trouble in mind: Paralinguistic indices of effort and uncertainty in communication. In Oralité et gestualité: Interactions et comportements multimodaux dans la communication. Edited by C. Cavé, I. Guaïtella, and S. Santi, 597–600. Paris: L’Harmattan.

                                      Save Citation »Export Citation »E-mail Citation »

                                      Uses a mouse-tracking paradigm in which listeners followed instructions to click on complex and novel objects, which were either previously mentioned (given) or not (new). Following a disfluent um, listeners were faster to identify a new referent. This finding paralleled a production experiment: speakers were disfluent more often when referring to something new.

                                      Find this resource:

                                      • Barr, D. J., and M. Seyfeddinipur. 2010. The role of fillers in listener attributions for speaker disfluency. Language and Cognitive Processes 25.4: 441–455.

                                        DOI: 10.1080/01690960903047122Save Citation »Export Citation »E-mail Citation »

                                        Uses a mouse-tracking paradigm to show that listeners expect a discourse-new referent following a disfluent um, but this expectation depended specifically on what was new or old to the speaker.

                                        Find this resource:

                                        • Corley, M., L. J. MacGregor, and D. I. Donaldson. 2007. It’s the way that you, er, say it: Hesitations in speech affect language comprehension. Cognition 105.3: 658–668.

                                          DOI: 10.1016/j.cognition.2006.10.010Save Citation »Export Citation »E-mail Citation »

                                          Shows that disfluency reduces the difficulty that is associated with processing unpredictable words; this finding is consistent with other work in this section showing that disfluency ameliorates processing of difficult information (See also Attention and Memory.)

                                          Find this resource:

                                          • Watanabe, M., K. Hirose, Y. Den, and N. Minematsu. 2008. Filled pauses as cues to the complexity of upcoming phrases for native and non-native listeners. Speech Communication 50.2: 81–94.

                                            DOI: 10.1016/j.specom.2007.06.002Save Citation »Export Citation »E-mail Citation »

                                            Reaction times were faster to identify a complex shape following a filled pause in Japanese both for native Japanese listeners and non-native Chinese listeners. In their task, the linguistic phrase and object were both complex, so either linguistic or conceptual complexity could have driven this effect.

                                            Find this resource:

                                            As a Cue to the Speaker’s Meaning: Children

                                            Studies in As a Cue to the Speaker’s Meaning: Adults demonstrate that adults can use disfluency in real time to direct their attention toward certain referents. Kidd, et al. 2011 examines this ability in very young children, showing that they also have a disfluency bias toward a referent that is both new and novel.

                                            Parsing and Its Timecourse

                                            Related to disfluency as a cue to the speaker’s meaning (see As a Cue to the Speaker’s Meaning: Adults and As a Cue to the Speaker’s Meaning: Children), a specific question is whether disfluency affects the processes of parsing—that is, identifying the syntactic structure of a spoken utterance. Ferreira and Bailey 2004 and Ferreira, et al. 2004 provide overviews of research, to date, as well as computational models. Several studies examined how disfluency and other temporal delays affect grammaticality judgments. Bailey and Ferreira 2003 and Bailey and Ferreira 2004 examine the parsing of subordinate/main-clause ambiguities and coordination ambiguities, while Lau and Ferreira 2005 examines main-verb/reduced-relative ambiguities. Maxfield, et al. 2009 uses ERPs to examine subordinate/main-clause ambiguities. On the other hand, Bailey and Ferreira 2007 finds that disfluency affects the interpretation of a referential ambiguity but not parsing.

                                            • Bailey, K. G. D., and F. Ferreira. 2003. Disfluencies affect the parsing of garden-path sentences. Journal of Memory and Language 49.2: 183–200.

                                              DOI: 10.1016/S0749-596X(03)00027-5Save Citation »Export Citation »E-mail Citation »

                                              Grammaticality judgment experiments demonstrate that both disfluencies and other non-speech interruptions affect the perception of an utterance as grammatical. Concludes that temporal delays from any interruption, including dog barks, affect parsing.

                                              Find this resource:

                                              • Bailey, K. G. D., and F. Ferreira. 2004. The disfluent hairy dog: Can syntactic parsing be affected by nonword disfluencies? In Approaches to studying world-situated language use: Bridging the language-as-product and language-as-action traditions. Edited by J. C. Trueswell and M. K. Tanenhaus, 303–316. Learning, Development, and Conceptual Change. Cambridge, MA: MIT Press.

                                                Save Citation »Export Citation »E-mail Citation »

                                                Reports preliminary data from the grammaticality judgment experiments in Bailey and Ferreira 2003 and discusses them as an approach to bridging the product and action traditions in psycholinguistics.

                                                Find this resource:

                                                • Bailey, K. G. D., and F. Ferreira. 2007. The processing of filled pause disfluencies in the visual world. In Eye movements: A window on mind and brain. Edited by R. P. G. van Gompel, M. H. Fischer, W. S. Murray, and R. L. Hill, 487–502. Educational Psychology. Oxford: Elsevier.

                                                  Save Citation »Export Citation »E-mail Citation »

                                                  Uses a visual-world eye-tracking paradigm to examine the effect of disfluencies immediately preceding referential ambiguities in sentences that also contained PP-attachment ambiguities (e.g., “Put the uh uh apple on the towel in the box”). In contrast to the other studies in this section, this study does not find an effect of disfluency on parsing, but instead only on the interpretation of the referential ambiguity.

                                                  Find this resource:

                                                  • Ferreira, F., and K. G. D. Bailey. 2004. Disfluencies and human language comprehension. Trends in Cognitive Sciences 8.5: 231–237.

                                                    DOI: 10.1016/j.tics.2004.03.011Save Citation »Export Citation »E-mail Citation »

                                                    Overview article; summarizes research on syntactic parsing and how disfluency affects it, and presents a computational model. This is a good article for an advanced undergraduate, a graduate, or a professional looking for an overview. Even though the title refers to language comprehension overall, the focus is specifically on parsing. (See also General Overviews.)

                                                    Find this resource:

                                                    • Ferreira, F., E. F. Lau, and K. G. D. Bailey. 2004. Disfluencies, language comprehension, and tree adjoining grammars. In 2003 Rumelhart Prize special issue honoring Aravind K. Joshi. Cognitive Science 28.5: 721–749.

                                                      DOI: 10.1016/j.cogsci.2003.10.006Save Citation »Export Citation »E-mail Citation »

                                                      Reviews research on how disfluency affects parsing, and presents a model of how disfluency affects processing, by using a tree adjoining grammar (TAG).

                                                      Find this resource:

                                                      • Lau, E. F., and F. Ferreira. 2005. Lingering effects of disfluent material on comprehension of garden path sentences. Language and Cognitive Processes 20.5: 633–666.

                                                        DOI: 10.1080/01690960444000142Save Citation »Export Citation »E-mail Citation »

                                                        Uses a grammaticality judgment task with auditory sentences to examine the comprehension of sentences with a main-verb/reduced-relative ambiguity. (See also As a Cue to Error Correction.)

                                                        Find this resource:

                                                        • Maxfield, N. D., J. M. Lyon, and E. R. Silliman. 2009. Disfluencies along the garden path: Brain electrophysiological evidence of disrupted sentence processing. Brain and Language 111.2: 86–100.

                                                          DOI: 10.1016/j.bandl.2009.08.003Save Citation »Export Citation »E-mail Citation »

                                                          Extends the findings in Bailey and Ferreira 2003 with ERP methods and discusses possible mechanisms for the effects of disfluencies on parsing garden path sentences.

                                                          Find this resource:

                                                          As a Cue to Error Correction

                                                          Disfluencies are commonly considered “mistakes,” but they are not the same thing as a speech error, which is defined as saying a different word or phrase than intended (see Speech Errors). Nevertheless, disfluencies are often associated with errors. For example, if a speaker stops and corrects an error, the result is a nonfluent sentence (e.g., “the car, uh, I mean the truck”). The studies in this section test questions about how repairs are processed. Brennan and Schober 2001 demonstrates that filled pauses (uh and um) can facilitate comprehension of corrected material, and Howell and Young 1991 shows that prosodic cues can facilitate correction processes. At the same time, several studies, including Corley 2010 and Lau and Ferreira 2005, show that repaired information is not entirely excised from the listener’s representations.

                                                          • Brennan, S. E., and M. F. Schober. 2001. How listeners compensate for disfluencies in spontaneous speech. Journal of Memory and Language 44.2: 274–296.

                                                            DOI: 10.1006/jmla.2000.2753Save Citation »Export Citation »E-mail Citation »

                                                            Tests the effect of disfluencies on listeners’ ability to accommodate corrections (e.g., in “Move to the yel-, uh, purple square”). Shows that filled pauses provide the time that listeners need to cancel misleading information and to help identify the correct object more quickly, suggesting that disfluency affects listeners’ expectations (see also Parsing and Its Timecourse). This study uses a naturalistic approach to studying comprehension, where spontaneously produced utterances were the stimuli for comprehension experiments.

                                                            Find this resource:

                                                            • Corley, M. 2010. Making predictions from speech with repairs: Evidence from eye movements. Language and Cognitive Processes 25.5: 706–727.

                                                              DOI: 10.1080/01690960903512489Save Citation »Export Citation »E-mail Citation »

                                                              Examines whether repaired information is overridden or is integrated into the final interpretation. Listeners’ eye movements were tracked as they listened to repaired statements such as “The boy will eat—uh, move the cake.” Results were mixed: later eye movements, but not eye movements at the repaired verb, suggested that repaired information was not completely overridden.

                                                              Find this resource:

                                                              • Howell, P., and K. Young. 1991. The use of prosody in highlighting alterations in repairs from unrestricted speech. In Special issue: Hearing and speech. Quarterly Journal of Experimental Psychology, Section A: Human Experimental Psychology 43.3: 733–758.

                                                                DOI: 10.1080/14640749108400994Save Citation »Export Citation »E-mail Citation »

                                                                Shows that prosodic cues can help listeners understand sentences in which the speaker corrects himself or herself.

                                                                Find this resource:

                                                                • Lau, E. F., and F. Ferreira. 2005. Lingering effects of disfluent material on comprehension of garden path sentences. Language and Cognitive Processes 20.5: 633–666.

                                                                  DOI: 10.1080/01690960444000142Save Citation »Export Citation »E-mail Citation »

                                                                  Uses a grammaticality judgment task with auditory sentences. Shows that repaired material is not completely overridden. Instead, proposes an “overlay” mechanism. (See also Parsing and Its Timecourse.)

                                                                  Find this resource:

                                                                  As a Cue to the Speaker’s Metacognitive State

                                                                  Brennan and Williams 1995 shows that disfluency can also be used to draw inferences about the speaker’s mental state; for example, whether the speaker is confident about the information being conveyed. Barr 2003 demonstrates that this has consequences both for language comprehension and learning. This information is related to questions about how disfluency leads to inferences about the message content, because metacognitive information can be used to draw inferences about the speaker’s meaning. (See also As a Cue to the Speaker’s Meaning: Adults.)

                                                                  • Barr, D. J. 2003. Paralinguistic correlates of conceptual structure. Psychonomic Bulletin & Review 10.2: 462–467.

                                                                    DOI: 10.3758/BF03196507Save Citation »Export Citation »E-mail Citation »

                                                                    Shows that disfluency carries information about the speaker’s certainty, and listeners use it while learning new concepts. Novel color category names were learned more quickly when the “teacher” spoke hesitantly for colors that were close to the boundary.

                                                                    Find this resource:

                                                                    • Brennan, S. E., and M. Williams. 1995. The feeling of another’s knowing: Prosody and filled pauses as cues to listeners about the metacognitive states of speakers. Journal of Memory and Language 34.3: 383–398.

                                                                      DOI: 10.1006/jmla.1995.1017Save Citation »Export Citation »E-mail Citation »

                                                                      Examines how listeners use prosody and disfluency to rate the speaker’s feeling of knowing.

                                                                      Find this resource:

                                                                      In Production

                                                                      The causes and characteristics of disfluent speech have been widely investigated. Under the assumption that many disfluencies result from disruptions in speech planning, many studies have tested the specific problems that result in disfluencies or in particular types of disfluencies. The section on Language Production summarizes these studies; work in the section on Nonlinguistic Influences surveys other variables, such as conversational roles and demographic characteristics, which may also influence when or how speakers become disfluent. One disfluency type that has received particular attention in theories of language production is repairs; several studies, covered in Timing of Repairs, have investigated when speakers detect the need for a repair, initiate the repair, and resume speech. Finally, work reviewed in Acoustics has sought to characterize the acoustic properties of disfluent speech so that disfluencies can be recognized or produced by computer speech technology.

                                                                      Language Production

                                                                      A substantial body of work has examined what processes cause speakers to become disfluent. Maclay and Osgood 1959 is an influential early investigation that demonstrates that disfluencies fall into several categories; Shriberg 1996 provides a more recent examination of the distribution of disfluency. Disfluency has frequently been viewed as stemming from difficulty or uncertainty in language production, as exemplified in Schachter, et al. 1991; a related argument, presented in Postma, et al. 1990, is that disfluency results from covert repairs of speech errors before they are produced. Subsequent work has often examined in detail the causes and functions of particular disfluencies: Clark and Fox Tree 2002 investigates fillers; Clark and Wasow 1998, repeated words; and Fox Tree and Clark 1997, prolonged words. Some work has considered whether filled pauses such as uh and um should be considered lexical items deliberately produced by speakers to signal a delay: Clark and Fox Tree 2002 presents an argument in favor of this hypothesis, and Finlayson and Corley 2012 provides an argument against.

                                                                      • Clark, H. H., and J. E. Fox Tree. 2002. Using uh and um in spontaneous speaking. Cognition 84.1: 73–111.

                                                                        DOI: 10.1016/S0010-0277(02)00017-3Save Citation »Export Citation »E-mail Citation »

                                                                        Provides evidence that filled pauses typically precede delays in speech and vary in form depending on the length of the coming delay. Proposes the influential hypothesis that speakers use filled pauses as an intentional signal to account for upcoming delays in speech production.

                                                                        Find this resource:

                                                                        • Clark, H. H., and T. Wasow. 1998. Repeating words in spontaneous speech. Cognitive Psychology 37.3: 201–242.

                                                                          DOI: 10.1006/cogp.1998.0693Save Citation »Export Citation »E-mail Citation »

                                                                          Comprehensive examination of disfluent repeated words. Argues they reflect a commit-and-repair strategy in which speakers begin an utterance before it is fully planned and then repeat the beginning of the utterance if the end is not fully planned in time.

                                                                          Find this resource:

                                                                          • Finlayson, I. R., and M. Corley. 2012. Disfluency in dialogue: An intentional signal from the speaker? Psychonomic Bulletin & Review 19.5: 921–928.

                                                                            DOI: 10.3758/s13423-012-0279-xSave Citation »Export Citation »E-mail Citation »

                                                                            Provides a counterargument against Clark and Fox Tree’s proposal that filled pauses are a deliberate linguistic signal.

                                                                            Find this resource:

                                                                            • Fox Tree, J. E., and H. H. Clark. 1997. Pronouncing “the” as “thee” to signal problems in speaking. Cognition 62.2: 151–167.

                                                                              DOI: 10.1016/S0010-0277(96)00781-0Save Citation »Export Citation »E-mail Citation »

                                                                              Examines a frequently overlooked category of disfluency: prolonged words.

                                                                              Find this resource:

                                                                              • Maclay, H., and C. E. Osgood. 1959. Hesitation phenomena in spontaneous English speech. Word 15.1: 19–44.

                                                                                Save Citation »Export Citation »E-mail Citation »

                                                                                Influential early investigation of disfluency. Establishes that speakers produce multiple types of disfluencies. Provides evidence that disfluencies are not randomly located in speech but instead reflect constraints on production processes.

                                                                                Find this resource:

                                                                                • Postma, A., H. Kolk, and D.-J. Povel. 1990. On the relation among speech errors, disfluencies, and self-repairs. Language and Speech 33.1: 19–29.

                                                                                  DOI: 10.1177/002383099003300102Save Citation »Export Citation »E-mail Citation »

                                                                                  Manipulates the emphasis placed on accuracy in a speech production task, finding that speakers produce more disfluencies under conditions emphasizing accuracy. Presents the influential hypothesis that disfluencies stem from covert repairs of yet-to-be-produced speech.

                                                                                  Find this resource:

                                                                                  • Schachter, S., N. Christenfeld, B. Ravina, and F. Bilous. 1991. Speech disfluency and the structure of knowledge. Journal of Personality and Social Psychology 60.3: 362–367.

                                                                                    DOI: 10.1037/0022-3514.60.3.362Save Citation »Export Citation »E-mail Citation »

                                                                                    Measures the rate of disfluency in college lectures on different topics. Provides naturalistic evidence that disfluency is associated with uncertainty in speech planning.

                                                                                    Find this resource:

                                                                                    • Shriberg, E. 1996. Disfluencies in Switchboard. In ICSLP 96 Proceedings: Fourth International Conference on Spoken Language Processing, October 3–6, 1996, Wyndham Franklin Plaza Hotel, Philadelphia, PA, USA. Vol. 3,. Edited by H. T. Bunnell and W. Idsardi, 11–14. Wilmington, DE: Alfred I. DuPont Institute, Applied Science and Engineering Laboratories.

                                                                                      Save Citation »Export Citation »E-mail Citation »

                                                                                      Provides data on the frequency and location of multiple categories of disfluency, on the basis of the large and widely used Switchboard corpus.

                                                                                      Find this resource:

                                                                                      Timing of Repairs

                                                                                      When speakers make an error, they often interrupt their speech to produce a disfluent repair of what they said previously. Levelt 1983 presents a breakdown of the process of repair into multiple subprocesses—interrupting speech, marking the repair, resuming speech—that has guided subsequent research. Although Levelt 1983 proposes that speakers interrupt their speech as soon as they detect an error, subsequent work has provided evidence against this work: Berg 1986 and Blackmer and Mitton 1991 do so by examining relatively unconstrained speech, while Tydgat, et al. 2011 provides similar arguments by experimentally manipulating when a repair becomes necessary. Oomen and Postma 2001 further shows that the timing of repair processes can accommodate differences in speech rate. Brown-Schmidt and Tanenhaus 2006 balances experimental control and naturalistic productions by using eye-tracking to track speakers’ free viewing of controlled visual displays. Finally, van Wijk and Kempen 1987 uses the timing of repairs as a source of evidence to argue that multiple cognitive processes are available to repair speech, depending on the changes that must be made.

                                                                                      • Berg, T. 1986. The aftermath of error occurrence: Psycholinguistic evidence from cut-offs. Language & Communication 6.3: 195–213.

                                                                                        DOI: 10.1016/0271-5309(86)90023-6Save Citation »Export Citation »E-mail Citation »

                                                                                        Counters the proposal in Levelt 1983. Argues that speakers do not always immediately interrupt themselves upon detecting an error, often waiting until the end of a word.

                                                                                        Find this resource:

                                                                                        • Blackmer, E. R., and J. L. Mitton. 1991. Theories of monitoring and the timing of repairs in spontaneous speech. Cognition 39.3: 173–194.

                                                                                          DOI: 10.1016/0010-0277(91)90052-6Save Citation »Export Citation »E-mail Citation »

                                                                                          Measures, in speakers’ free productions, the delay between when speakers interrupt speech and when they resume speaking. The observed short delays provide important evidence that speakers must have planned part of the repair before the interruption and do not always interrupt speech immediately upon detecting an error.

                                                                                          Find this resource:

                                                                                          • Brown-Schmidt, S., and M. K. Tanenhaus. 2006. Watching the eyes when talking about size: An investigation of message formulation and utterance planning. Journal of Memory and Language 54.4: 592–609.

                                                                                            DOI: 10.1016/j.jml.2005.12.008Save Citation »Export Citation »E-mail Citation »

                                                                                            Uses eye tracking of speakers’ viewing of a scene to track when speakers first become aware of information that might require a revision of a prior description. Demonstrates that this timing influences whether a fluent utterance, filled pause, or repair is produced. Demonstrates how eye tracking can be used to study disfluency in language production.

                                                                                            Find this resource:

                                                                                            • Levelt, W. J. M. 1983. Monitoring and self-repair in speech. Cognition 14.1: 41–104.

                                                                                              DOI: 10.1016/0010-0277(83)90026-4Save Citation »Export Citation »E-mail Citation »

                                                                                              Breaks down the process of repair to highlight that researchers must consider when speakers interrupt their speech, how they mark their interruption, and how they resume. Argues that speakers may be delayed in detecting their errors but interrupt speech as soon as they detect the error.

                                                                                              Find this resource:

                                                                                              • Oomen, C. C. E., and A. Postma. 2001. Effects of time pressure on mechanisms of speech production and self-monitoring. Journal of Psycholinguistic Research 30.2: 163–184.

                                                                                                DOI: 10.1023/A:1010377828778Save Citation »Export Citation »E-mail Citation »

                                                                                                Uses a task in which participants adjust their speech rate to the speed of a moving dot. Speakers who must speak quickly interrupt and repair themselves faster but are no less accurate in doing so. Provides evidence that monitoring one’s language production does not require a fixed length of time.

                                                                                                Find this resource:

                                                                                                • Tydgat, I., M. Stevens, R. J. Hartsuiker, and M. J. Pickering. 2011. Deciding where to stop speaking. Journal of Memory and Language 64.4: 359–380.

                                                                                                  DOI: 10.1016/j.jml.2011.02.002Save Citation »Export Citation »E-mail Citation »

                                                                                                  Experimentally manipulates changes in a visual display that speakers are asked to describe, to test when speakers will interrupt their speech once they have realized that a repair is necessary. Provides a current look at the self-repair literature.

                                                                                                  Find this resource:

                                                                                                  • van Wijk, C., and G. Kempen. 1987. A dual system for producing self-repairs in spontaneous speech: Evidence from experimentally elicited corrections. Cognitive Psychology 19.4: 403–440.

                                                                                                    DOI: 10.1016/0010-0285(87)90014-4Save Citation »Export Citation »E-mail Citation »

                                                                                                    Uses data on where speakers interrupt and restart speech, to argue that repair can involve two different processes depending on whether the utterance must be completely reformulated or new words can simply be substituted into an existing frame. Highlights that self-repairs may not be a unitary phenomenon.

                                                                                                    Find this resource:

                                                                                                    Nonlinguistic Influences

                                                                                                    Although disfluencies are often attributed to difficulty in language production, as described in Language Production, some work has examined their relation to nonlinguistic factors. Some of this work, rather than considering what conditions lead particular speakers to become disfluent, has examined how disfluency rates vary across speakers. Bortfeld, et al. 2001 provides a comprehensive examination of multiple demographic characteristics, while Schmid and Fägersten 2010 examines bilinguals. In addition, since most speech occurs in a conversation, it is important to examine the influence of the conversational context, such as who is leading the conversation; Branigan, et al. 1999 provides such an investigation. Naive theories of speech may often attribute disfluency to nervousness or anxiety; the authors of Reynolds and Paivio 1968 tested this hypothesis but found only a limited influence of anxiety on the rate of disfluency. Naive theories may also lead lying speakers to omit fillers, in an effort to enhance their perceived credibility. This hypothesis was tested in Arciuli, et al. 2010, which examines fillers in lies versus truthful statements. Although filled pauses and repetitions are often viewed as symptoms of production problems that speakers would have preferred to avoid, other work, exemplified in Schegloff 2010, argues that disfluencies may in fact sometimes help get things done in conversation. This work has applied the technique of conversational analysis, by examining the sequences of turns that interlocutors produce in a conversation. Finally, Swerts and Krahmer 2005 establishes facial cues that tend to accompany disfluency in language production.

                                                                                                    • Arciuli, J., D. Mallard, and G. Villar. 2010. “Um, I can tell you’re lying”: Linguistic markers of deception versus truth-telling in speech. Applied Psycholinguistics 31.3: 397–411.

                                                                                                      DOI: 10.1017/S0142716410000044Save Citation »Export Citation »E-mail Citation »

                                                                                                      Demonstrates that speakers telling lies use fewer filled pauses than when telling the truth. Results are of applied interest.

                                                                                                      Find this resource:

                                                                                                      • Bortfeld, H., S. D. Leon, J. E. Bloom, M. F. Schober, and S. E. Brennan. 2001. Disfluency rates in conversation: Effects of age, relationship, topic, role, and gender. Language and Speech 44.2: 123–147.

                                                                                                        DOI: 10.1177/00238309010440020101Save Citation »Export Citation »E-mail Citation »

                                                                                                        Establishes the influence of multiple demographic characteristics both of speakers and listeners on disfluency.

                                                                                                        Find this resource:

                                                                                                        • Branigan, H. P., R. Lickley, and D. McKelvie. 1999. “Non-linguistic influences on rates of disfluency in spontaneous speech.” In Proceedings of the 14th International Congress of Phonetic Sciences: ICPhS 99, San Francisco, 1–7 August 1999. Edited by J. J. Ohala, Y. Hasegawa, M. Ohala, D. Granville, and A. C. Bailey, 387–390. Berkeley: Linguistics Department, Univ. of California.

                                                                                                          Save Citation »Export Citation »E-mail Citation »

                                                                                                          Provides an investigation of how conversational factors, such as the ability to make eye contact, affect disfluency rates.

                                                                                                          Find this resource:

                                                                                                          • Reynolds, A., and A. Paivio. 1968. Cognitive and emotional determinants of speech. Canadian Journal of Psychology 22.3: 164–175.

                                                                                                            DOI: 10.1037/h0082757Save Citation »Export Citation »E-mail Citation »

                                                                                                            Evaluates the influence of anxiety on disfluency production. One of only a small number of studies to examine noncognitive influences on disfluency.

                                                                                                            Find this resource:

                                                                                                            • Schegloff, E. A. 2010. Some other “uh(m)”s. Discourse Processes 47.2: 130–174.

                                                                                                              DOI: 10.1080/01638530903223380Save Citation »Export Citation »E-mail Citation »

                                                                                                              Argues against the common assumption that uh and um always reflect delay in the process of speech production. For example, a person making a phone call may use um to mark a shift from initial greetings to the reason for calling. Demonstrates conversational analytic approach.

                                                                                                              Find this resource:

                                                                                                              • Schmid, M. S., and K. B. Fägersten. 2010. Disfluency markers in L1 attrition. Language Learning 60.4: 753–791.

                                                                                                                DOI: 10.1111/j.1467-9922.2010.00575.xSave Citation »Export Citation »E-mail Citation »

                                                                                                                Examines the nature and causes of an increase in disfluency with disuse of a speaker’s native language. Provides an important investigation of disfluency in the context of bilingualism.

                                                                                                                Find this resource:

                                                                                                                • Swerts, M., and E. Krahmer. 2005. Audiovisual prosody and feeling of knowing. Journal of Memory and Language 53.1: 81–94.

                                                                                                                  DOI: 10.1016/j.jml.2005.02.003Save Citation »Export Citation »E-mail Citation »

                                                                                                                  Demonstrates that particular facial cues accompany disfluency in speech production and that comprehenders can use these cues to infer an uncertain speaker even in the absence of speech.

                                                                                                                  Find this resource:

                                                                                                                  Acoustics

                                                                                                                  Work has examined the acoustic characteristics of speech before, during, and after a disfluent interruption, frequently with the goal of facilitating computer-based automatic speech recognition or synthesis. Shriberg 1994; Shriberg 2001; and Bell, et al. 2003 provide large-scale analyses of speech corpora. Although these studies do not exclusively examine the acoustics of disfluencies, they are notable for the scale both of the corpora examined and the data reported. Duez 1985 and Swerts 1998 give more-focused acoustic analyses of silent pauses and filled pauses, respectively.

                                                                                                                  • Bell, A., D. Jurafsky, E. Fosler-Lussier, C. Girand, M. Gregory, and D. Gildea. 2003. Effects of disfluencies, predictability, and utterance position on word form variation in English conversation. Journal of the Acoustical Society of America 113.2: 1001–1024.

                                                                                                                    DOI: 10.1121/1.1534836Save Citation »Export Citation »E-mail Citation »

                                                                                                                    Large-scale corpus investigation of function words. Although the overall goal of the paper is to explain variation in duration and lenition in production, a substantial portion of the work investigates the consequences of disfluency for how these words are realized. Stands out for its large corpus and detailed statistical models.

                                                                                                                    Find this resource:

                                                                                                                    • Duez, D. 1985. Perception of silent pauses in continuous speech. Language and Speech 28.4: 377–389.

                                                                                                                      DOI: 10.1177/002383098502800403Save Citation »Export Citation »E-mail Citation »

                                                                                                                      Provides a detailed investigation of which acoustic, semantic, and syntactic cues contribute to the perception that a silent pause is disfluent.

                                                                                                                      Find this resource:

                                                                                                                      • Shriberg, E. 1994. Preliminaries to a theory of speech disfluencies. PhD diss., University of California, Berkeley.

                                                                                                                        Save Citation »Export Citation »E-mail Citation »

                                                                                                                        Analyzes several corpora both of person-to-person and person-to-computer speech. Contains an extremely comprehensive report on form and frequency of multiple types of disfluency, followed by analyses of some of the acoustic characteristics (pitch and duration) of these disfluencies. Also noteworthy for including person-to-machine speech, relevant to automatic speech-recognition applications.

                                                                                                                        Find this resource:

                                                                                                                        • Shriberg, E. 2001. To “errrr” is human: Ecology and acoustics of speech disfluencies. Journal of the International Phonetic Association 31.1: 153–169.

                                                                                                                          DOI: 10.1017/S0025100301001128Save Citation »Export Citation »E-mail Citation »

                                                                                                                          Provides extensive data on the frequency and location of multiple disfluency types, reporting detailed examinations of duration and voice quality of multiple types of disfluency. Further breaks down data by several regions of an individual disfluency (error, editing phrase, and repair), providing a comprehensive examination of the acoustics of disfluency.

                                                                                                                          Find this resource:

                                                                                                                          • Swerts, M. 1998. Filled pauses as markers of discourse structure. Journal of Pragmatics 30.4: 485–496.

                                                                                                                            DOI: 10.1016/S0378-2166(98)00014-9Save Citation »Export Citation »E-mail Citation »

                                                                                                                            Investigates the characteristics of filled pauses in elicited monologues. Establishes the acoustic characteristics (pitch and duration) of filled pauses and how these cues vary as a function of their location in speech.

                                                                                                                            Find this resource:

                                                                                                                            In Development

                                                                                                                            Language learners both produce and comprehend disfluent speech. Studies reviewed in Children’s Production have investigated how and why children produce disfluent speech, to assess development in typically developing children or to establish baselines for comparison to fluency disorders. In turn, comprehending disfluent speech might be particularly difficult for the language learner, but studies reviewed in Children’s Comprehension indicate that disfluent speech need not disrupt and sometimes may even provide cues to language acquisition and referential processing. Finally, studies cited under Second-Language Acquisition assess whether disfluency when speaking a second language can be used as an index of proficiency in that language.

                                                                                                                            Children’s Production

                                                                                                                            Like adults, children are frequently disfluent in their speech. Because of clinical interest in childhood stuttering, a number of studies have examined non-stuttering children’s production of disfluencies, with the aim of establishing control data; Yaruss, et al. 1999 presents one such example. Other work has investigated non-stuttering children’s disfluencies in their own right. Rispoli and Hadley 2001 and Rispoli, et al. 2008 examine how changes in children’s processing ability with age influences their fluency. Perry and Lewis 1999 uses children’s disfluencies to measure their conceptual learning. Together, these works indicate that disfluency is not always a consequence of poor cognitive or linguistic performance.

                                                                                                                            • Perry, M., and J. L. Lewis. 1999. Verbal imprecision as an index of knowledge in transition. Developmental Psychology 35.3: 749–759.

                                                                                                                              DOI: 10.1037/0012-1649.35.3.749Save Citation »Export Citation »E-mail Citation »

                                                                                                                              Examines young children’s speech as they explain solutions to physics problems. Reports that children who better learn the task exhibit more disfluencies, which is attributed to searches for relevant knowledge. Provides important evidence that disfluencies do not necessarily indicate poor task performance, and demonstrates how disfluencies can be used to study the development of knowledge within a domain.

                                                                                                                              Find this resource:

                                                                                                                              • Rispoli, M., and P. Hadley. 2001. The leading-edge: The significance of sentence disruptions in the development of grammar. Journal of Speech, Language, and Hearing Research 44.5: 1131–1143.

                                                                                                                                DOI: 10.1044/1092-4388(2001/089)Save Citation »Export Citation »E-mail Citation »

                                                                                                                                Demonstrates that young children’s disfluent utterances are longer and more complicated than their fluent utterances, proposing that these utterances were disfluent because they were beyond the children’s current processing ability in language production. Demonstrates how the processing limitations of young children can contribute to disfluency.

                                                                                                                                Find this resource:

                                                                                                                                • Rispoli, M., P. Hadley, and J. Holt. 2008. Stalls and revisions: A developmental perspective on sentence production. Journal of Speech, Language, and Hearing Research 51.4: 953–966.

                                                                                                                                  DOI: 10.1044/1092-4388(2008/070)Save Citation »Export Citation »E-mail Citation »

                                                                                                                                  Finds that revisions (or repairs) increase with age, while other disfluency types do not. Proposes that this is because revisions require additional processes, such as self-monitoring, that are slower to develop. Valuable for providing a longitudinal examination of how the fluency of the same children’s productions changes with age.

                                                                                                                                  Find this resource:

                                                                                                                                  • Yaruss, J. S., R. M. Newman, and T. Flora. 1999. Language and disfluency in nonstuttering children’s conversational speech. Journal of Fluency Disorders 24.3: 185–207.

                                                                                                                                    DOI: 10.1016/S0094-730X(99)00009-1Save Citation »Export Citation »E-mail Citation »

                                                                                                                                    Examines disfluency in children’s spontaneous speech. Reports data on children’s fluency, with the goal of establishing norms for comparison to stuttering children. This study also makes the important methodological contributions of contrasting within-participant correlations (e.g., between utterance length and fluency) with between-participant correlations.

                                                                                                                                    Find this resource:

                                                                                                                                    Children’s Comprehension

                                                                                                                                    A potential challenge in language acquisition is that the input is often disfluent. Soderstrom and Morgan 2007 provides an outline of how this complicates language acquisition, as well as experimental evidence supporting a possible solution: children preferentially attend to fluent speech. Kidd, et al. 2011 presents evidence that disfluency is not necessarily harmful for comprehension; children, like adults, can use a speaker’s disfluency to predict what he or she will refer to.

                                                                                                                                    Second-Language Acquisition

                                                                                                                                    Disfluency has also been proposed as a measure of second-language proficiency. Segalowitz 2010 emphasizes the need to distinguish among cognitive disfluency, perceived speech disfluency, and actual speech disfluency in the second language. In longitudinal studies, Lennon 1990 reports that decreased disfluency predicts gains in ratings of fluency by trained instructors, and Segalowitz and Freed 2004 concludes that decreased disfluency predicts gains in measures of underlying cognitive processes. Riazantseva 2001 reports similar results, by using a cross-sectional design. However, Riggenbach 1991 and Kormos and Dénes 2004 demonstrate that such relations are not always observed and that there is variability among listeners in the cues used to assess fluency. One confounding factor may be speech style: Riazantseva 2001 suggests that languages differ in their overall speaking or pausing style, and de Jong, et al. 2013 demonstrates that these measures can be improved as predictors of second-language fluency by controlling for such differences. Finally, Kormos 1999 reviews evidence about when and how second-language speakers correct and repair errors in their speech.

                                                                                                                                    • de Jong, N. H., R. Groenhout, R. Schoonen, and J. H. Hulstijn. 2013. Second language fluency: Speaking style or proficiency? Correcting measures of second language fluency for first language behavior. Applied Psycholinguistics.

                                                                                                                                      DOI: 10.1017/S0142716413000210Save Citation »Export Citation »E-mail Citation »

                                                                                                                                      Demonstrates that pause-related measures of second-language fluency can be improved by correcting for overall speech rate, which may reflect differences among individuals or languages in speaking style.

                                                                                                                                      Find this resource:

                                                                                                                                      • Kormos, J. 1999. Monitoring and self-repair in L2. Language Learning 49.2: 303–342.

                                                                                                                                        DOI: 10.1111/0023-8333.00090Save Citation »Export Citation »E-mail Citation »

                                                                                                                                        Reviews how and when speakers can detect and correct errors in their second language. Concludes that second-language speakers’ repairs target errors of grammatical and phonological form more than errors of message-level content, as compared to native speakers’ repairs. Further concludes that more-proficient second-language speakers repair more of their errors.

                                                                                                                                        Find this resource:

                                                                                                                                        • Kormos, J., and M. Dénes. 2004. Exploring measures and perceptions of fluency in the speech of second language learners. System 32.2: 145–164.

                                                                                                                                          DOI: 10.1016/j.system.2004.01.001Save Citation »Export Citation »E-mail Citation »

                                                                                                                                          Uses a cross-sectional design. Examines which disfluencies predict fluency as predicted both by native and non-native listeners. Finds that disfluencies are generally not predictive of perceived fluency and that there is variability across listeners in the cues used to assess fluency.

                                                                                                                                          Find this resource:

                                                                                                                                          • Lennon, P. 1990. Investigating fluency in EFL: A quantitative approach. Language Learning 40.3: 387–417.

                                                                                                                                            DOI: 10.1111/j.1467-1770.1990.tb00669.xSave Citation »Export Citation »E-mail Citation »

                                                                                                                                            Uses a longitudinal design to determine which decreases in which kind of disfluencies predict gains in ratings of fluency by trained instructors. Fewer filled pauses and repetitions predicted such gains, but repairs did not. Demonstrates that some kinds of disfluencies in a second language are related to perceived fluency.

                                                                                                                                            Find this resource:

                                                                                                                                            • Riazantseva, A. 2001. Second language proficiency and pausing: A study of Russian speakers of English. Studies in Second Language Acquisition 23.4: 497–526.

                                                                                                                                              DOI: 10.1017/S027226310100403XSave Citation »Export Citation »E-mail Citation »

                                                                                                                                              Uses a cross-sectional design to demonstrate that silent pauses differ across second-language learner groups, defined by their scores on standard measures of proficiency. Additionally provides evidence for cross-linguistic differences in pausing even in native-language speech.

                                                                                                                                              Find this resource:

                                                                                                                                              • Riggenbach, H. 1991. Toward an understanding of fluency: A microanalysis of nonnative speaker conversations. Discourse Processes 14.4: 423–441.

                                                                                                                                                DOI: 10.1080/01638539109544795Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                Uses a cross-sectional design to compare which disfluencies in second-language speakers’ dialogues predict ratings of fluency by their instructors. Finds that disfluencies generally do not differ across groups. Also examines conversational factors such as backchannel responses.

                                                                                                                                                Find this resource:

                                                                                                                                                • Segalowitz, N. 2010. Cognitive bases of second language fluency. Cognitive Sciences and Second Language Acquisition. New York: Routledge.

                                                                                                                                                  Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                  Emphasizes the need to distinguish between underlying cognitive fluency and the fluency of spoken speech and between objective disfluency and disfluency as perceived by listeners.

                                                                                                                                                  Find this resource:

                                                                                                                                                  • Segalowitz, N., and B. F. Freed. 2004. Context, contact, and cognition in oral fluency acquisition: Learning Spanish in at home and study abroad contexts. In Special issue: Learning context and its effects on second language acquisition. Studies in Second Language Acquisition 26.2: 173–199.

                                                                                                                                                    DOI: 10.1017/S0272263104262027Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                    Uses a longitudinal design to determine which kind of disfluencies predict gains in measures of cognitive processing in a second language, such as judging whether a word refers to an animate object. Demonstrates that disfluencies in a second language are also related to more-objective measures of cognitive processing.

                                                                                                                                                    Find this resource:

                                                                                                                                                    In Computational Linguistics

                                                                                                                                                    Disfluency has attracted considerable attention within the field of computational linguistics because it poses a challenge both for speech recognition systems and text-to-speech systems. In this section, some representative articles that report on the role of disfluency in speech recognition, speech synthesis, and human-computer interaction are cited. Brennan 2000 provides a good overview discussion of disfluency and why it is relevant to computational linguistics.

                                                                                                                                                    • Brennan, S. E. 2000. Processes that shape conversation and their implications for computational linguistics. In 38th Annual Meeting of the Association for Computational Linguistics: Proceedings of the conference, 1–8 October 2000, Hong Kong. Edited by O. Kwong, 1–11. San Francisco: Morgan Kaufmann.

                                                                                                                                                      Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                      This overview article provides a concise summary of disfluency phenomena. It draws on work from psycholinguistics and conversation analysis and makes the argument that disfluency research is important for computational linguistics and human-computer interaction.

                                                                                                                                                      Find this resource:

                                                                                                                                                      Speech Recognition

                                                                                                                                                      Disfluency poses a challenge to speech recognition systems, yet a successful system should be able to deal with naturalistic speech input, including disfluencies. A representative sample of computational models is cited here, with papers organized around four topics: (1) methods of detecting disfluency overall, including multiple types, (2) methods of specifically detecting repairs, (3) the effect of disfluency on automatic word recognition, and (4) the role of disfluency in computer speech and human-computer interaction.

                                                                                                                                                      Detecting Multiple Types of Disfluency

                                                                                                                                                      Papers in this section propose computational models for identifying disfluency in the speech input. The model in Stolcke and Shriberg 1996 predicts disfluency and uses it to help predict upcoming words; the model in Stolcke, et al. 1998 detects both sentence boundaries and disfluencies. Models differ in the types of information used to identify disfluencies: the model in Shriberg, et al. 1997 uses only local prosodic cues to detect disfluencies, while the model in Snover, et al. 2004 depends primarily on lexical features. While most of these models are designed to detect English disfluencies, work on other languages is presented in Rodríguez and Torres 2006 (for Spanish) and Stouten, et al. 2006 (for Dutch).

                                                                                                                                                      • Rodríguez, L. J., and M. I. Torres. 2006. Spontaneous speech events in two speech databases of human-computer and human-human dialogs in Spanish. Language and Speech 49.3: 333–366.

                                                                                                                                                        DOI: 10.1177/00238309060490030201Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                        Discusses the benefits of modeling disfluencies and other spontaneous speech events in models of speech recognition for the Spanish language.

                                                                                                                                                        Find this resource:

                                                                                                                                                        • Shriberg, E., R. Bates, and A. Stolcke. 1997. A prosody-only decision-tree model for disfluency detection. In Proceedings: Eurospeech ’97, 5th European Conference on Speech Communication and Technology, Rhodes, Greece, 22–25 September 1997. Vol. 5. Edited by G. Kokkinakis, 2383–2386. Grenoble, France: ESCA.

                                                                                                                                                          Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                          Presents a computational model that shows disfluencies can be detected in a model that uses only local prosodic cues.

                                                                                                                                                          Find this resource:

                                                                                                                                                          • Snover, M., B. Dorr, and R. Schwartz. 2004. A lexically-driven algorithm for disfluency detection. In HLT-NAACL 2004: Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics, May 2–7, 2004, Boston, Massachusetts, USA; Short papers. Edited by J. Hirschberg, 157–160. East Stroudsburg, PA: Association for Computational Linguistics.

                                                                                                                                                            Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                            Presents a different kind of model than the others in this section, in that it attempts to identify disfluencies by using just lexical features (words and part-of-speech labels), without depending on prosodic cues.

                                                                                                                                                            Find this resource:

                                                                                                                                                            • Stolcke, A., and E. Shriberg. 1996. Statistical language modeling for speech disfluencies. In The 1996 IEEE International Conference on Acoustics, Speech, and Signal Processing: Conference proceedings, May 7–10, 1996, Marriott Marquis Hotel, Atlanta, Georgia, USA. Vol. 1, Speech (Part 1). 405–409. Piscataway, NJ: Institute of Electrical and Electronics Engineers Service Center.

                                                                                                                                                              Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                              Presents an N-gram-style statistical language model that predicts disfluency, and creates an edited fluent utterance that the model uses to predict the following word in an utterance.

                                                                                                                                                              Find this resource:

                                                                                                                                                              • Stolcke, A., E. Shriberg, R. Bates, et al. 1998. Automatic detection of sentence boundaries and disfluencies based on recognized words. In ICSLP ’98: Proceedings of the 5th International Conference on Spoken Language Processing, incorporating the Seventh Australian International Speech and Technology Conference held at the Sydney Convention Centre, 30 November–4 December 1998. Vol. 5. Edited by R. H. Mannell and J. Robert-Ribes, 2247–2250. Sydney, Australia: Australian Speech Science and Technology Association.

                                                                                                                                                                Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                Provides a computational model for detecting sentence boundaries and disfluencies. Uses both a prosodic model and an n-gram event language model.

                                                                                                                                                                Find this resource:

                                                                                                                                                                • Stouten, F., J. Duchateau, J.-P. Martens, and P. Wambacq. 2006. Coping with disfluencies in spontaneous speech recognition: Acoustic detection and linguistic context manipulation. In Special issue: Robustness issues for conversational interaction. Speech Communication 48.11: 1590–1606.

                                                                                                                                                                  DOI: 10.1016/j.specom.2006.04.004Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                  Proposes methods to deal with filled pauses, repeated words, and sentence restarts in a computational speech recognition model. The model is tested both against the Switchboard corpus of American English and the CGN corpus of Dutch.

                                                                                                                                                                  Find this resource:

                                                                                                                                                                  Detecting Repairs

                                                                                                                                                                  One important role of disfluency is to help signal corrections. Shriberg, et al. 1992 analyzes the characteristics of repairs produced by humans interacting with a computer; Nakatani and Hirschberg 1994 and Heeman and Allen 1994 present computational models for automatic repair detection.

                                                                                                                                                                  • Heeman, P. A., and J. Allen. 1994. Detecting and correcting speech repairs. In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics, 27–30 June 1994, New Mexico State University, Las Cruces, New Mexico, USA. Edited by J. Pustejovsky, 295–302. Morristown, NJ: Association for Computational Linguistics.

                                                                                                                                                                    Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                    Proposes that computational models of human dialogue would benefit from simultaneously doing part-of-speech tagging and identifying intonational phrases, speech repairs, and discourse markers.

                                                                                                                                                                    Find this resource:

                                                                                                                                                                    • Nakatani, C. H., and J. Hirschberg. 1994. A corpus-based study of repair cues in spontaneous speech. Journal of the Acoustical Society of America 95.3: 1603–1616.

                                                                                                                                                                      DOI: 10.1121/1.408547Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                      Proposes a model that builds on the authors’ earlier model of repair interval, using acoustic-prosodic cues to repair detection.

                                                                                                                                                                      Find this resource:

                                                                                                                                                                      • Shriberg, E., J. Bear, and J. Dowding. 1992. Automatic detection and correction of repairs in human-computer dialog. Paper presented at the Fifth DARPA Speech and Natural Language Workshop, held at the Arden House Conference Center, Harriman, NY, in February 1992. In Proceedings of the DARPA Speech and Natural Language Workshop. Edited by M. P. Marcus, 419–424. San Francisco: Morgan Kaufmann.

                                                                                                                                                                        Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                        Presents an analysis of repairs in a human-computer speech corpus and identifies criteria for detecting repairs by using semantic, syntactic, and acoustic information. These criteria are tested in experiments using the Gemini natural-language processing system. A critical property of this system is that it does not rely on an explicit edit signal.

                                                                                                                                                                        Find this resource:

                                                                                                                                                                        Automatic Word Recognition

                                                                                                                                                                        Goldwater, et al. 2010 examines the effect of disfluency on automatic word recognition; for discussion of how disfluency affects human word recognition, see Effects on Speech Comprehension: Word Recognition.

                                                                                                                                                                        • Goldwater, S., D. Jurafsky, and C. D. Manning. 2010. Which words are hard to recognize? Prosodic, lexical, and disfluency factors that increase speech recognition error rates. Speech Communication 52.3: 181–200.

                                                                                                                                                                          DOI: 10.1016/j.specom.2009.10.001Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                          Presents the first systematic examination of the effect of disfluency on the success of recognizing a word. Examines two automatic speech recognition systems. Demonstrates that predictors of errors include a disfluent interruption point, extreme prosody, and a “doubly confusable pair” in the lexicon. One strength of this paper is that it examines disfluency together with lexical and prosodic characteristics of speech and with speaker differences.

                                                                                                                                                                          Find this resource:

                                                                                                                                                                          Production in Computer Systems and Human-Computer interaction

                                                                                                                                                                          Adell, et al. 2012 discusses how disfluency can be incorporated in text-to-speech systems, and Oviatt 1995 presents an analysis of human speech in the context of human-computer interaction. Ehlen, et al. 2007 uses human disfluency to improve automatic survey systems.

                                                                                                                                                                          • Adell, J., D. Escudero, and A. Bonafonte. 2012. Production of filled pauses in concatenative speech synthesis based on the underlying fluent sentence. Speech Communication 54.3: 459–476.

                                                                                                                                                                            DOI: 10.1016/j.specom.2011.10.010Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                            Argues that text-to-speech systems would benefit from simulating human speech styles, including disfluencies, and proposes a model to do so. This moves beyond the traditional focus on reading styles in text-to-speech systems.

                                                                                                                                                                            Find this resource:

                                                                                                                                                                            • Ehlen, P., M. F. Schober, and F. G. Conrad. 2007. Modeling speech disfluency to predict conceptual misalignment in speech survey interfaces. Discourse Processes 44.3: 245–265.

                                                                                                                                                                              DOI: 10.1080/01638530701600839Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                              Evaluates disfluency as a metric for automatically identifying situations when there is conceptual misalignment in automatic computer survey systems.

                                                                                                                                                                              Find this resource:

                                                                                                                                                                              • Oviatt, S. 1995. Predicting spoken disfluencies during human-computer interaction. Computer Speech & Language 9.1: 19–35.

                                                                                                                                                                                DOI: 10.1006/csla.1995.0002Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                Provides a characterization of the kinds of disfluencies that occur during human-computer interaction, both in speech and writing, including proportion of filled pauses (speech only), content corrections, spelling and abbreviations (writing only), repetitions, and false starts. An experiment finds that disfluency is higher in an unconstrained task than in a structured one.

                                                                                                                                                                                Find this resource:

                                                                                                                                                                                Clinical Research

                                                                                                                                                                                Although disfluency in typical language production is not identical to stuttering, studies have often sought to relate the two phenomena; Postma and Kolk 1993 provides a representative and influential example of such work. Some other work has examined how disfluencies are produced by speakers with other clinical diagnoses: Engelhardt, et al. 2010 and Engelhardt, et al. 2011 examine attention-deficit/hyperactivity disorder (ADHD), and Lake, et al. 2011 and Lunsford, et al. 2010 examine autism spectrum disorders (ASD). In addition to its clinical application, examining how disfluencies are produced in speakers with differing inhibitory or perspective-taking abilities can provide insight into the mechanisms underlying disfluency in the language production system.

                                                                                                                                                                                • Engelhardt, P. E., M. Corley, J. T. Nigg, and F. Ferreira. 2010. The role of inhibition in the production of disfluencies. Memory & Cognition 38.5: 617–628.

                                                                                                                                                                                  DOI: 10.3758/MC.38.5.617Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                  Reports that speakers with ADHD produce more repairs—but not other types of disfluency—because they might not sufficiently plan ahead and thus need to correct what they said. Establishes that different disfluency types can be differently affected by particular disorders. Supports the idea that different disfluencies can represent distinct problems in language production.

                                                                                                                                                                                  Find this resource:

                                                                                                                                                                                  • Engelhardt, P. E., F. Ferreira, and J. T. Nigg. 2011. Language production strategies and disfluencies in multi-clause network descriptions: A study of adult attention-deficit/hyperactivity disorder. Neuropsychology 25.4: 442–453.

                                                                                                                                                                                    DOI: 10.1037/a0022436Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                    Unlike the authors’ prior study, finds that speakers with ADHD produce more silent pauses and repetitions rather than repairs. Attributes the different findings to task differences, which suggests that the task chosen may be an important methodological consideration in clinical examinations of disfluency.

                                                                                                                                                                                    Find this resource:

                                                                                                                                                                                    • Lake, J. K., K. R. Humphreys, and S. Cardy. 2011. Listener vs. speaker-oriented aspects of speech: Studying the disfluencies of individuals with autism spectrum disorders. Psychonomic Bulletin & Review 18.1: 135–140.

                                                                                                                                                                                      DOI: 10.3758/s13423-010-0037-xSave Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                      Expands the range of clinical diagnoses related to disfluency, by examining speakers with ASD. Demonstrates that speakers with ASD, who may lack a “theory of mind,” produce fewer fillers and repetitions. Argues this pattern is consistent with previous proposals that fillers are a listener-oriented signal. Demonstrates how clinical data can inform broader hypotheses about disfluency.

                                                                                                                                                                                      Find this resource:

                                                                                                                                                                                      • Lunsford, R., P. A. Heeman, L. Black, and J. van Santen. 2010. Autism and the use of fillers: Differences between “um” and “uh.” In Proceedings of DiSS-LPSS Joint Workshop 2010, the 5th Workshop on Disfluency in Spontaneous Speech and the 2nd International Symposium on Linguistic Patterns in Spontaneous Speech, 25–26 September 2010, Tokyo, Japan. Edited by K. Maekawa and Y. Den, 107–110. Tokyo: University of Tokyo.

                                                                                                                                                                                        Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                        Examines the production of uh and um in dialogues between clinicians and children with ASD and in typical development (TD) controls. They find that children with ASD use fewer ums than TD children but do not differ in their rate of producing the filler uh.

                                                                                                                                                                                        Find this resource:

                                                                                                                                                                                        • Postma, A., and H. Kolk. 1993. The covert repair hypothesis: Prearticulatory repair processes in normal and stuttered disfluencies. Journal of Speech and Hearing Research 36.3: 472–487.

                                                                                                                                                                                          Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                          Influential examination of how disfluency relates to stuttering. On the basis of the hypothesis that disfluency represents a covert repair of an error in speech that was yet to be articulated, argues that stutterers produce more disfluencies simply because they have more errors to be corrected.

                                                                                                                                                                                          Find this resource:

                                                                                                                                                                                          Speech Errors

                                                                                                                                                                                          The term “disfluency” is most commonly applied to hesitations and interruptions in the flow of speech. Disfluencies are typically distinguished from speech errors, in which speakers choose the wrong word or mispronounce a word without necessarily interrupting the fluent flow of speech; speech errors may sometimes lead to a disfluent repair (see Timing of Repairs), but not necessarily. Classic studies of speech errors, including Fromkin 1971, Garrett 1980, and Dell and Reich 1981, established that speech errors are not arbitrarily distributed in speech and that their distribution can reveal the stages and representations involved in speech production. Dell 1986 implements a connectionist model of language production on the basis of these and other data. Other work has examined how speakers monitor their own speech for speech errors and attempt to avoid them; Baars, et al. 1975 represents a classic investigation on this topic, and Hartsuiker, et al. 2005 provides a modern reply. Finally, speech errors have been used as evidence to address other questions in language production; Oppenheim and Dell 2008 uses speech errors to examine representations in overtly articulated speech versus inner speech.

                                                                                                                                                                                          • Baars, B. J., M. T. Motley, and D. G. MacKay. 1975. Output editing for lexical status in artificially elicited slips of the tongue. Journal of Verbal Learning and Verbal Behavior 14.4: 382–391.

                                                                                                                                                                                            DOI: 10.1016/S0022-5371(75)80017-XSave Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                            Exemplifies a widely used experimental procedure to elicit a large number of speech errors for study, as well as how this task can be experimentally manipulated. Demonstrates that speech errors are more frequent when the slipped-to word is an actual word, and discusses the implications for monitoring processes.

                                                                                                                                                                                            Find this resource:

                                                                                                                                                                                            • Dell, G. S. 1986. A spreading-activation theory of retrieval in sentence production. Psychological Review 93.3: 283–321.

                                                                                                                                                                                              DOI: 10.1037/0033-295X.93.3.283Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                              Implements a successful and widely cited connectionist model of language production. Demonstrates how speech errors can be modeled and used to constrain theories of language production.

                                                                                                                                                                                              Find this resource:

                                                                                                                                                                                              • Dell, G. S., and P. A. Reich. 1981. Stages in sentence production: An analysis of speech error data. Journal of Verbal Learning and Verbal Behavior 20.6: 611–629.

                                                                                                                                                                                                DOI: 10.1016/S0022-5371(81)90202-4Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                                Establishes that speech errors related both semantically and phonologically to the intended word are more common than would be expected from separate, additive effects, indicating interactive processes. Provides both an important theoretical development in the study of language production and a demonstration of how speech error corpora can be analyzed.

                                                                                                                                                                                                Find this resource:

                                                                                                                                                                                                • Fromkin, V. A. 1971. The non-anomalous nature of anomalous utterances. Language 47.1: 27–52.

                                                                                                                                                                                                  DOI: 10.2307/412187Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                                  Important early work in establishing that speech errors are not randomly distributed and that they can reflect constraints on speech production.

                                                                                                                                                                                                  Find this resource:

                                                                                                                                                                                                  • Garrett, M. F. 1980. Levels of processing in sentence production. In Language production. Vol. 1, Speech and talk. Edited by B. Butterworth, 177–220. Orlando, FL, and London: Academic Press.

                                                                                                                                                                                                    Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                                    Establishes that speech errors typically result in syntactically and phonologically well-formed utterances in which some elements are simply misplaced. Demonstrates that word and sound errors occur over differently sized domains. Extremely influential both in establishing characteristics of typical speech errors and in using them to support multistage models of language production.

                                                                                                                                                                                                    Find this resource:

                                                                                                                                                                                                    • Hartsuiker, R. J., M. Corley, and H. Martensen. 2005. The lexical bias effect is modulated by context, but the standard monitoring account doesn’t fly: Related beply to Baars et al. (1975). Journal of Memory and Language 52.1: 58–70.

                                                                                                                                                                                                      DOI: 10.1016/j.jml.2004.07.006Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                                      Provides an updated look at self-monitoring of speech errors, with a different interpretation than in Baars, et al. 1975.

                                                                                                                                                                                                      Find this resource:

                                                                                                                                                                                                      • Oppenheim, G. M., and G. S. Dell. 2008. Inner speech slips exhibit lexical bias, but not the phonemic similarity effect. Cognition 106.1: 528–537.

                                                                                                                                                                                                        DOI: 10.1016/j.cognition.2007.02.006Save Citation »Export Citation »E-mail Citation »

                                                                                                                                                                                                        Contrasts speech errors made in inner speech versus overtly articulated speech while producing tongue-twisters. Provides an example of early-21st-century research using speech errors and demonstrates how they can be used to investigate the representation of speech.

                                                                                                                                                                                                        Find this resource:

                                                                                                                                                                                                        back to top

                                                                                                                                                                                                        Article

                                                                                                                                                                                                        Up

                                                                                                                                                                                                        Down