Hyperarticulation of Corrections in Multilingual Dialogue Systems
The DOI matches its Crossref record; 19 of the 19 references it cites were found in a registry or an index.
Abstract
This present paper aims at answering the question whether there are distinctive cross-linguistic differences associated with hyperarticulated speech in correction dialogue acts. The objective is to assess the effort for adaptation of a multilingual dialogue system in 9 different languages, regarding the recovery strategies, particularly corrections. If the presence of hyperarticulation significantly differs across languages, it will have a significant impact on the dialogue design and recovery strategies.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Certain domains are aligned with particular platform types, indicating specialised applications e.g. the smart home assistance domain was associated with devices such as Amazon Alexa [6, 17, 19, 33, 46, 60], the INSPIRE system [74], and devices not specified [38].”
From System and User Strategies to Repair Conversational Breakdowns of Spoken Dialogue Systems: A Scoping Review (2024) · cites this paper as [38]
In the citing paper
“By contrast, studies within the customer service domain utilised telephones [9, 11, 15, 27, 32, 66] or computer speakers [29, 38].”
From System and User Strategies to Repair Conversational Breakdowns of Spoken Dialogue Systems: A Scoping Review (2024) · cites this paper as [38]
In the citing paper
“Sample size ranged from as few as five participants in a qualitative study in a real-world setting [60] to as many as 850 participants in an online multi-language quantitative study [38].”
From System and User Strategies to Repair Conversational Breakdowns of Spoken Dialogue Systems: A Scoping Review (2024) · cites this paper as [38]
In the citing paper
“All but one study employed single-language SDSs, the exception being [38], which implemented multilingual SDSs across various regions to study differences in user responses to system breakdowns in different linguistic contexts.”
From System and User Strategies to Repair Conversational Breakdowns of Spoken Dialogue Systems: A Scoping Review (2024) · cites this paper as [38]
In the citing paper
“Similarly, [38] found that speakers' responses to communication breakdowns varied significantly by language, with 55.3% of Italian speakers but only 9.6% of Turkish speakers using a special keyword as a systemcentric repair strategy.”
From System and User Strategies to Repair Conversational Breakdowns of Spoken Dialogue Systems: A Scoping Review (2024) · cites this paper as [38]
Citations
Cites
Fitting Linear Mixed-Effects Models Using lme4
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015) · Journal of Statistical Software · doi:10.18637/jss.v067.i01 · Cited by 86,835 · cited 1×
“Since the prosody features were measured on speech recorded in a complex experimental setup with speakers in many languages, better insight about the effects of the factors could be given by employing Linear Mixed Models [24].”
MethodPraat, a system for doing phonetics by computer
Boersma, P. (2002) · UvA-DARE (University of Amsterdam) · Cited by 4,417
Praat script to detect syllable nuclei and measure speech rate automatically
de Jong, N. H., & Wempe, T. (2009) · Behavior Research Methods · doi:10.3758/brm.41.2.385 · Cited by 596
ProsodyPro — A Tool for Large-scale Systematic Prosody Analysis
Xu, Y. (2013) · UCL Discovery (University College London) · Cited by 317
Cross-language differences in fundamental frequency range: A comparison of English and German
Mennen, I., Schaeffler, F., & Docherty, G. (2012) · The Journal of the Acoustical Society of America · doi:10.1121/1.3681950 · Cited by 150 · cited 1×
“Although there are many research studies dealing with cross-linguistic prosodic differences [16-19], they are mostly done on a pair of languages and on a limited number of participants.”
BackgroundProsodic and other cues to speech recognition failures
Hirschberg, J., Litman, D. J., & Swerts, M. G. J. (2004) · Speech Communication · doi:10.1016/j.specom.2004.01.006 · Cited by 106 · cited 2×
“In [7] the duration, pause, and pitch features were employed to train a decision tree classifier, which was extended and integrated with recognizer confidence scores for further improvements in detection of corrections [8].”
BackgroundMethodAltenberg, E. P., & Ferrand, C. T. (2005) · Journal of Voice · doi:10.1016/j.jvoice.2005.01.005 · Cited by 100 · cited 1×
“Although there are many research studies dealing with cross-linguistic prosodic differences [16-19], they are mostly done on a pair of languages and on a limited number of participants.”
BackgroundAdapting speaking after evidence of misrecognition: Local and global hyperarticulation
Stent, A. J., Huffman, M. K., & Brennan, S. E. (2007) · Speech Communication · doi:10.1016/j.specom.2007.07.005 · Cited by 85 · cited 2×
“However other studies reported minor changes [15] or no negative impact [13] on error rate.”
BackgroundCorrections in spoken dialogue systems
Swerts, M., Litman, D., & Hirschberg, J. (2000) · 6th International Conference on Spoken Language Processing (ICSLP 2000) · doi:10.21437/icslp.2000-344 · Cited by 81 · cited 2×
“Firstly, the presence of characteristic hyperarticulated speech for correction turns has to be confirmed, in [9-10] it is defined, with reference to prosody, as “slower and louder speech with wider pitch excursion and more internal silence”.”
BackgroundCharacterizing and recognizing spoken corrections in human-computer dialogue
Levow, G.-A. (1998) · Proceedings of the 36th annual meeting on Association for Computational Linguistics - · doi:10.3115/980845.980969 · Cited by 80 · cited 1×
“In [7] the duration, pause, and pitch features were employed to train a decision tree classifier, which was extended and integrated with recognizer confidence scores for further improvements in detection of corrections [8].”
Method
Cited by
Alghamdi, E., Halvey, M., & Nicol, E. (2024) · ACM Conversational User Interfaces 2024 · doi:10.1145/3640794.3665558 · Cited by 36
“Almost all studies included in the review were conducted in a single language, except one [38], which was a multi-language study encompassing nine languages.”
BackgroundResultInfluentialError correction and extraction in request dialogs
Constantin, S., & Waibel, A. (2022) · Zenodo (CERN European Organization for Nuclear Research) · doi:10.5281/zenodo.7924937 · Cited by 1
Cross-Lingual Transfer Learning for Affective Spoken Dialogue Systems
Gjoreski, K., Gjoreski, A., Kraljevski, I., & Hirschfeld, D. (2019) · Interspeech (USB) · doi:10.21437/interspeech.2019-2163 · Cited by 1
“In our previous work [9], cross-lingual differences related to hyperarticulated speech in correction dialogue acts were investigated.”
BackgroundSelf-citationError correction and extraction in request dialogsPreprint
Constantin, S., & Waibel, A. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.04243 · Cited by 1
“Kraljevski and Hirschfeld (2017) propose a domain-independent correction detection by checking the speech for hyperarticulation.”
Comparison of Error Correction and Extraction Approaches
Constantin, S., & Waibel, A. (2023) · Signals and communication technology · doi:10.1007/978-3-031-44260-5_5
Error correction and extraction in request dialogs
Constantin, S., & Waibel, A. (2020) · Repository KITopen (Karlsruhe Institute of Technology) · doi:10.5445/ir/1000158071
Error correction and extraction in request dialogs
Constantin, S., & Waibel, A. (2022) · Repository KITopen (Karlsruhe Institute of Technology) · doi:10.5445/ir/1000158055
Constantin, S. G. (2026) · Repository KITopen (Karlsruhe Institute of Technology) · doi:10.5445/ir/1000193274
Metadata
- Authors
- Kraljevski, I., & Hirschfeld, D.
- ORCID
- Kraljevski, Ivan
- Venue
- Interspeech 2017
- Publisher
- ISCA
- Published
- 2017-08-20
- Pages
- 2531-2535
- ISSN
- 2308-457X (ISSN-L)
- OpenAlex
- W2750398874
- Cited by
- 3
- References
- 19
- Citations by year
- 2024: 1 · 2023: 1 · 2019: 1
- FWCI
- 0.12
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Interspeech (USB) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Social Sciences › Arts and Humanities › Language and Linguistics › Language, Discourse, Communication Strategies
- Keywords
- hyperarticulation, speech correction, multilingual dialogue systems, speech recognition
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.