Consistency of Judgments of Articulatory Productions

Irwin, Ruth Beckey

doi:10.1044/jshr.1303.548

Cited by 6 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Indeed, /r/ is notoriously difficult to code reliably, even among trained phoneticians; Lawson et al (2014) coded Glaswegian English /r/ into seven categories, and allowing for one category leeway (e.g., counting 'no /r/' and 'derhotic' as agreement), the three authors achieved 84-86% agreement. Comparable findings for other variables have been reported by Irwin (1970) (86-87% test-retest consistency for labeling consonant misarticulations); Pitt, Johnson, Hume, Kiesling, and Raymond (2005) (92.9% inter-coder agreement for stops, 86.5% for liquids); Fosler-Lussier, Dilley, Tyson, and Pitt (2007) (79.4-80.9% inter-coder agreement for stops, 74.7-79.0% for liquids/glides); and Hall-Lew and Fix (2012) (by-token SDs around 0.8-1.0 for coding /l/ vocalization on a 4-point scale). Hall-Lew and Fix ( 2012) also note that the 'intermediate' tokens (those with mean ratings in the middle of the scale) experienced the least reliable ratings.…”

Section: Discussionsupporting

confidence: 82%

From categories to gradience: Auto-coding sociophonetic variation with random forests

Villarreal

Clark

Hay

et al. 2020

Laboratory Phonology: Journal of the Association for Laboratory Phonology

View full text Add to dashboard Cite

The time-consuming nature of coding sociophonetic variables that are typically treated as categorical represents an impediment to addressing research questions around these variables that require large volumes of data. In this paper, we apply a machine learning method, random forest classification (Breiman, 2001), to automate coding (categorical prediction) of two English sociophonetic variables traditionally treated as categorical, non-prevocalic /r/ and word-medial intervocalic /t/, based on tokens' acoustic signatures. We found good performance for binary classifiers of non-prevocalic /r/ (Absent versus Present) and medial /t/ (Voiced versus Voiceless), but not for medial /t/ with a six-way coding distinction (largely due to some codes being sparsely represented in the training data). This method also yields rankings of acoustic measures in terms of importance in classification. Beyond any individual measures, this method generates probabilistic predictions of variation (classifier probabilities) that represent a composite of the acoustic cues fed into the model. In a listening experiment, we found that not only did classifier probabilities significantly capture gradience in trained listeners' perceptions of rhoticity, they better predicted listeners' perceptions than individual acoustic measures. This method thus represents a new approach to reconciling the categorical and continuous dimensions of sociophonetic variation.

show abstract

Section: Discussionsupporting

confidence: 82%

From categories to gradience: Auto-coding sociophonetic variation with random forests

Villarreal

Clark

Hay

et al. 2020

Laboratory Phonology: Journal of the Association for Laboratory Phonology

View full text Add to dashboard Cite

show abstract

“…See Kiesling et al ͑2006͒ for more details. 3 These labeling consistency data compare favorably with other studies ͑e.g., Irwin, 1970;Eisen, 1991͒. For example, Eisen ͑1991͒ found labeling accuracy of 88% for obstruent consonants.…”

Section: Discussionmentioning

confidence: 96%

“…A more recent unpublished test of intertranscriber reliability using eight labelers and 1 min of speech allowed us to specifically investigate agreement among canonical, deleted, and glottal variants. Agreement for these variants was 85.2%, indicating high reliability in line with previous findings of good interrater agreement ͑e.g., Irwin, 1970;Eisen, 1991͒. 3 The speech of 19 talkers ͑9 male, 10 female; approximately 138 000 words͒ was used to identify lexical sequences constituting assimilable environments, i.e., twoword sequences in which the place of articulation of the word-final phoneme could assimilate to that of a following word-initial phoneme.…”

Section: A Methodsmentioning

confidence: 99%

A study of regressive place assimilation in spontaneous speech and its implications for spoken word recognition

Dilley

Pitt

2007

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

Regressive place assimilation is a form of pronunciation variation in which a word-final alveolar sound takes the place of articulation of a following labial or velar sound, as when green boat is pronounced greem boat. How listeners recover the intended word (e.g., green, given greem) has been a major focus of spoken word recognition theories. However, the extent to which this variation occurs in casual, unscripted speech has previously not been reported. Two studies of pronunciation variation were conducted using a spontaneous speech corpus. First, phonetic labeling data were used to identify contexts in which assimilation could occur, namely, when a word-final alveolar stop (/t/, /d/, or /n/) was followed by a velar or labial consonant. Assimilation was indicated relatively infrequently, while deletion, glottalization, or canonical pronunciations were more often indicated. Moreover, lexical frequency was shown to affect pronunciation; high frequency lexical items showed more types of variation. Second, acoustic analyses showed that neither place of articulation cues (indicated by second formant variation) nor relative amplitude was sufficient to distinguish assimilated from deleted and canonical variants; only when closure duration was additionally taken into account were these three variant types distinguishable. Implications for theories of word recognition are discussed.

show abstract

“…Dating back to Henderson (1938), agreement on two-way decisions (correct versus incorrect) has been shown to be higher than on five-way scoring (correct, deletion, substitution, distortion, addition), presumably due to the increased complexity of the decision process and the lack of systematic response definitions for substitutions, additions, and distortions (Irwin, 1970;Irwin and Krafchick, 1965;Norris et al, 1980;Philips and Bzoch, 1969). Even when agreement in the present study is based on the lenient agreement criterion of any diacritic to mark a distortion, it averages only 48%.…”

Section: T J Y E Systems and Agreement Criteriamentioning

confidence: 96%

“…Irwin, 1970;Pye et al, 1988) when transcribed using broad phonetic transcription. Using the same criteria, only 3 of the 24 consonants (13%) and 9 of the 17 vowels (53 YO) had acceptable agreement when transcribed using narrow phonetic transcription.…”

Section: Consonant and Vowel Agreementmentioning

confidence: 99%

Reliability studies in broad and narrow phonetic transcription

Shriberg

Lof

1991

Clinical Linguistics & Phonetics

227

141

View full text Add to dashboard Cite

A 16-category framework is proposed to review the sources of variance in studies of phonetic transcription reliability. The same framework is used to analyse transcription agreement data collected in the course of a project in child phonology, including 22 reliability estimates from five consensus transcription teams who transcribed eight subject groups. Detailed agreement data at the level of consonants, vowels and diphthongs, feature classes, and diacritics are presented for each of the 16 categories, including such traditional measurement variables as sampling mode (continuous speech; articulation tests), agreement type (intrajudge; inter-judge), and level of transcription (broad; narrow). Tabular and plotted data are deliberately presented at the lowest feasible levels for readers interested in specific questions at these levels. A total of 16 generalizations about transcription reliability are derived from descriptive and inferential statistical findings. The primary conclusion is that for certain clinical and research tasks in communicative disorders, broad phonetic transcription appears to be reliable, whereas narrow transcription may be unreliable.A historical overview of transcription studies in communicative disorders in America suggests a division into two periods, beginning with Henderson's (1938) call for programmatic research. From the 1930s through approximately 1970, research reflected the emerging discipline's attempt to develop effective ways to rate, score, and transcribe deviant speech. Impelled by efficiency criteria the assessment paradigm evolved to a form of nominal scoring of responses evoked primarily by articulation test stimuli (e.g. Burkowsky). A second period, beginning in the early 1970s and putatively concluding with the decade of the 1980s, witnessed renewed interest in phonetic transcription as input for linguistic description procedures developed for young normally developing children and children with intelligibility deficits (e.g.

show abstract

Consistency of Judgments of Articulatory Productions

Cited by 6 publications

References 0 publications

From categories to gradience: Auto-coding sociophonetic variation with random forests

From categories to gradience: Auto-coding sociophonetic variation with random forests

A study of regressive place assimilation in spontaneous speech and its implications for spoken word recognition

Reliability studies in broad and narrow phonetic transcription

Contact Info

Product

Resources

About