Identifying unfamiliar voices: Examining the system variables of sample duration and parade size.

Voice identification parades can be unreliable due to the error-prone nature of earwitness responses. UK government guidelines recommend that voice parades should have nine voices, each played for 60 s. This makes parades resource-consuming to construct. In this article, we conducted two experiments to see if voice parade procedures could be simplified. In Experiment 1 (N = 271, 135 female), we investigated if reducing the duration of the voice samples on a nine-voice parade would negatively affect identification performance using both conventional logistic and signal detection approaches. In Experiment 2 (N = 270, 136 female), we first explored if the same sample duration conditions used in Experiment 1 would lead to different outcomes if we reduced the parade size to include only six voices. Following this, we pooled the data from both experiments to investigate the influence of target-position effects. The results show that 15-s sample durations result in statistically equivalent voice identification performance to the longer 60-s sample durations, but that the 30-s sample duration suffers in terms of overall signal sensitivity. This pattern of results was replicated using both a nine- and a six-voice parade. Performance on target-absent parades were at chance levels in both parade sizes, and response criteria were mostly liberal. In addition, unwanted position effects were present. The results provide initial evidence that the sample duration used in a voice parade may be reduced, but we argue that the guidelines recommending a parade with nine voices should be maintained to provide additional protection for a potentially innocent suspect given the low target-absent accuracy.

Telephone Transmission and Earwitnesses: Performance on Voice Parades Controlled for Voice Similarity.

The effect of telephone transmission on a listener's ability to recognise a speaker in a voice parade is investigated. A hundred listeners (25 per condition) heard 1 of 5 'target' voices, then returned a week later for a voice parade. The 4 conditions were: target exposure and parade both at studio quality; exposure and parade both at telephone quality; studio exposure with telephone parade, and vice versa. Fewer correct identifications followed from telephone exposure and parade (64%) than from studio exposure and parade (76%). Fewer still resulted for studio exposure/telephone parade (60%) and, dramatically, only 32% for telephone exposure/studio parade. Certain speakers were identified more readily than others across all conditions. Confidence ratings reflected this effect of speaker, but not the effect of exposure/parade condition.

Speech rhythm: a metaphor?

Is speech rhythmic? In the absence of evidence for a traditional view that languages strive to coordinate either syllables or stress-feet with regular time intervals, we consider the alternative that languages exhibit contrastive rhythm subsisting merely in the alternation of stronger and weaker elements. This is initially plausible, particularly for languages with a steep 'prominence gradient', i.e. a large disparity between stronger and weaker elements; but we point out that alternation is poorly achieved even by a 'stress-timed' language such as English, and, historically, languages have conspicuously failed to adopt simple phonological remedies that would ensure alternation. Languages seem more concerned to allow 'syntagmatic contrast' between successive units and to use durational effects to support linguistic functions than to facilitate rhythm. Furthermore, some languages (e.g. Tamil, Korean) lack the lexical prominence which would most straightforwardly underpin prominence of alternation. We conclude that speech is not incontestibly rhythmic, and may even be antirhythmic. However, its linguistic structure and patterning allow the metaphorical extension of rhythm in varying degrees and in different ways depending on the language, and it is this analogical process which allows speech to be matched to external rhythms.

The role of pitch and timing cues in the perception of phrasal grouping in Seoul Korean.

This paper reports two experiments in which listeners detected prosodic group boundaries in Seoul Korean speech, investigating how pitch and timing cues collaborate or compete with each other. Two types of timing cues, compensatory lengthening and group-final lengthening, were employed. The results show that both pitch and timing have demarcative functions which are exploited by listeners. However, listeners relied more on timing than pitch and this may be because in the experiments the pitch contour variations were limited to a small number of phonological categories, whereas temporal variations were more gradient. In addition, group-final lengthening was a more robust cue to the prosodic boundary than compensatory lengthening, and the integration of pitch and timing cues seems to be context-dependent. The results highlight the significance of local information in the universal grouping process and that the precise nature of the cues affect the way they integrate in perception.

The Pairwise Variability Index and coexisting rhythms in language.

The Pairwise Variability Index (PVI) has been widely used as a metric for quantifying rhythm in languages, often with a view to placing them on a continuum between notional categories of stress-timing and syllable-timing. We review the history of and rationale for the PVI, and point out three potential anomalies in the way the PVI has been applied. Following up one of these we apply the PVI to the level of the foot, and argue that stress-timing and syllable-timing are not points at either end of a continuum but orthogonal dimensions, so that a language can be (for instance) both syllable-timed and stress-timed. Results from Estonian, English, Mexican Spanish, and Castilian Spanish are presented which give some support for this view.

The analysis of low accentuation in Estonian.

In Estonian, as in a number of other languages, the nuclear pitch accent is often low and level. This paper presents two studies of this phenomenon. The first, a phonetic analysis of carefully structured read sentences shows that low accentuation can also spread to the prenuclear accents in an intonational phrase. The resulting sentence contours are used as evidence to evaluate alternative phonological analyses of low accentuation, and H + L* is shown to account best for the data. The second study presents quantitative evidence from fundamental frequency values which supports this phonological analysis. Finally, the distribution of prenuclear pitch accents is discussed. High and low accents can co-occur in an intonational phrase, but only in patterns obeying a specific sequential constraint. A fragment of an intonational grammar for Estonian is presented capturing the observed distributional restrictions.