George D. Allen

PHONOLOGICAL RHYTHM: DEFINITION AND DEVELOPMENT

Publisher Summary This chapter discusses the phonological rhythm. Rhythm is a form of structure that plays an organizing role in speech. Despite this all-inclusiveness of potential rhythmic sources in speech, some processes play a more dominant role than others in defining what is commonly felt to be the rhythm of a phrase. Speech communication is a motor–perceptual process, and thus, the rhythmic structures typically found in it are of the sort reasonable for speakers to produce and for listeners to perceive. The organizing rhythm of the spoken utterance can be influenced by virtually any element of speech that is subject to sequential patterning. Although children can be sensitive to some differences in prosodic contour as early as 18 months, their ability to produce consistent differences in stress-accent develops much later. Children become sensitive to the stress patterns of phrases at about the age of 2 years; however, they can begin to signal some accentual distinctions in their own speech soon after. Both their perception and their production of the full system of stress distinctions can remain inaccurate in some respects until the age of 12. Childrens early polysyllabic utterances show a high frequency of unreduced syllables.

Extensions to the International Phonetic Alphabet for the transcription of atypical speech

This paper introduces and illustrates the Extensions to the International Phonetic Alphabet (IPA) that have been recommended for the narrow transcription of disordered speech. The relationship between these Extensions and previous suggestions for transcribing atypical speech made by the Working Party for the Phonetic Representation of Disordered Speech (PRDS) is described. By including the transcription of aspects of connected speech such as voice quality, rate and intensity, and by permitting uncertainty in transcription to be indicated, the International Phonetic Association has sanctioned significant developments in transcription conventions that will be of use to many people concerned with the narrow transcription of speech.

Acoustic Level and Vocal Effort as Cues for the Loudness of Speech

A review of previous studies of speech loudness shows great variability in the derived psychophysical functions relating loudness to speech power [i.e., sound‐pressure level (SPL)] and other physical and psychological measures, such as subglottal pressure (SGP) and vocal effort. This paper argues that, because of the complex nature of speech production and perception, traditional scaling procedures that yield exponential relations between loudness and some other measure should be replaced by multidimensional techniques. Two experiments on speech loudness scaling using partial correlation analysis demonstrate that loudness judgments depend upon both acoustic cues, as measured by peak SPL, and vocal effort cues, as measured by peak SGP in one experiment and subjective effort in the other. The relative dependencies upon these cues are different for different listeners and are at least somewhat resistant to distortion via signal attenuation and/or masking with white noise. These differences between listeners ...

How the young French child avoids the pre-voicing problem for word-initial voiced stops

Macken & Barton (1980) noted an asymmetry in their young Spanish-learning childrens word-initial stop productions, as compared with English-learning childrens, namely, ‘the Spanish-learning children acquire “lead” voicing…after age four’, whereas ‘English-learning children acquire “long lag” stops…by about 2;6’. ‘One possible explanation’, they propose, ‘is that lead voicing is inherently difficult to produce or at least more difficult to learn than long lag…’. The purpose of this Note is to present data suggesting that young children learning French, which also employs phonetically pre-voiced stops for its phonemically voiced category, also show an asymmetry in their word-initial stop productions. Furthermore, this asymmetry supports the notion, proposed by Macken & Barton, that lead voicing is somehow ‘more difficult’ for these young children to learn to produce. However, the actual ‘strategy’ used by the French-learning children to circumvent this presumed difficulty is different from that used by Macken & Bartons Spanish-learning children. This difference supports the notion that children have a certain degree of latitude in finding their way through the maze of psycholinguistic development (Kiparsky & Menn 1977).

Some Suprasegmental Contours in French Two-Year-Old Children’s Speech

Six 2-year-old French (Aix-en-Provence) children were tape-recorded as they identified 46 pictures with polysyllabic names (e.g., elephant, tracteur). Approximately 200 utterances were then analyzed for fundamental frequency (Fo) contours over the word, peak intensity (I) differences between the last two syllables, and absolute and relative nuclear vowel durations (D) between the last two syllables. The principal findings were: (1) falling Fo contours outnumbered rising by about 2.5: 1; (2) rising Fo contours were somewhat narrower, on the average, than falling; (3) nearly half of the I contours were essentially level, and 84% involved differences of 6 dB or less; (4) the final vowel was, on the average, longer than the penultimate vowel by a ratio of about 1.6:1, or an absolute difference of about 70 ms; (5) Fo and I were moderately correlated, with wide rising and falling Fo being associated with wide rising and falling I contours, respectively; (6) there was a similar, though statistically weaker, relationship between Fo and D contours; and (7) there was essentially no statistical relationship between I and D contours. These findings show that French 2-year-olds are influenced by the prosodic shape of their parents’ language yet are also under the control of presumably universal speech production constraints.

Vowel duration measurement: A reliability study

In recent studies involving the correlation of speech segment durations, the measurement error in locating segment boundaries has usually been ignored or assumed to be constant and small. Since the relative magnitude of the error in locating such a boundary may vary greatly depending upon its phonetic context and the procedures employed, thereby affecting the derived intersegment covariances profoundly, an experiment was designed to estimate these errors directly. Several tokens of four sentences containing syllables differing in the quality, length, degree of stress, and consonantal context of their nuclear vowels were copied onto two‐track tapes and sent to several professional phoneticians with instructions to measure the durations of the indicated vowels, on first one track and then, at least two weeks later, the other. The resulting measurement errors varied systematically in several ways: (1) long, stressed vowels had greater errors than short, unstressed ones; (2) initial voiced stops (/b/) were ea...

Temporal Structure in Speech Production

Temporal sequential dependencies in speech can be realized as negative correlations between the variations in duration of successive articulations; for example, if the consonant in a consonant‐vowel syllable is articulated overlong, the following vowel will often be overshort, making the syllable less variable in duration than the sum of the variabilities of its parts. Temporal dependencies in an articulatory interval may be measured by the intervals relative variance (RV), i.e., the variance of the intervals duration divided by its length. Under an hypothesis of no sequential dependency between articulations, RV is approximately constant for all intervals. As sequential dependencies increase, RV decreases; units of articulation then emerge as speech segments with low RV. RV also separates two kinds of dependencies. Since RV, unlike relative error, is sensitive to changes of scale, tempo variations in repetitions of an utterance will artificially inflate the RV of tempo‐related intervals (e.g., rhythmic...

Interactive prosody training workstation.

Few devices exist to aid in the training of pitch, intensity, and rhythm of speech. The Interactive prosody training workstation (PW) employs state‐of‐the‐art technology to assist users in achieving clinician‐ or client‐programmable targets in each of these features. Two training interfaces are currently implemented. The simpler displays F0 and/or intensity in real time as a fluctuating one‐ or two‐dimensional display. Smoothing can be adjusted to accommodate varying levels of voice variations. With the more advanced interface, model utterances are presented using stored LPC‐coded speech. The user’s response is then compared to the model, using any of a wide variety of scoring methods. The user’s response may be played, in comparison to the model, as often as desired. A demonstration tape will be played showing one hearing impaired individual with a cochlear implant using the PW first to find his appropriate F0 register and then practicing pitch gestures appropriate for speech. Strong carryover of learnin...

Acoustic‐phonetic analysis of normal, loud, and Lombard speech in simulated cockpit conditions

It has long been recognized that raising ones voice causes perceptible changes in the speech signal beyond mere increase in overall intensity. This paper reports the results of an extensive analysis of speech produced in a simulated fighter cockpit environment, with helmet and oxygen mask in place. Eight talkers each produced 56 utterances under three conditions: (1) normal, (2) loud (nominally 10 dB above normal), and (3) Lombard (evoked by 90 dB of pink noise played through a headset). A total of 17 671 phonemes were hand marked for analysis using 18 acoustic features (ten frequency bands, spectral COG, low‐ and high‐frequency spectral tilt, F0, F1–3, and duration). For most speakers, both loud and Lombard conditions showed the following shifts in comparison with the normal condition: (1) For vowels and sonorant consonants, lower (0–500 Hz) and higher (5–8 kHz) frequency bands lost energy relative to the mid (1–4 kHz) frequencies; (2) also for vowels and sonorants, F0, F1, and spectral COG all rose; an...

Temporal Variability of Repeated Utterances by Normal Speakers and Stutterers

Ten normal speakers, five stutterers, and five “cured stutterers” each repeated four different speech samples several times. The four samples were (1) an “easy” sentence, (2) a “hard” sentence, (3) an “easy” paragraph, and (4) a “hard” paragraph, where “easy” and “hard” refer to high and low frequency of occurrence in English of the content words used. In a separate fifth task, each subject tapped his finger as steadily as he could for 4 min. From each of these five repeated behaviors will be derived a variability measure (relative variance, or amount of variance per unit of time) that describes each subjects accuracy in repeating the time program of that behavior. The reliability of this measure, i.e., the stability of the measure across tasks for each subject will be examined, and its validity will be related to the normal and stuttering populations tested.