4.5. Linguistic Errors
5.2.2. Articulator Discoordination
The classic context in which impairment in AS for combining intact primary functions is demonstrated is provided by voice
onset time (VOT) studies. Speech dyspraxics, except some who are initially mute for speech (speculatively, unable to coordinate breath control and vocal cord adduction, and/or integrate the neuromuscular programming to 'set* the larynx for phonation), have no difficulty phonating. Neither do they necessarily evidence problems achieving bilabial, alveolar etc closure for stop consonant production. They commonly, though, show apparent difficulty in upholding normal voiced-voiceless distinctions.
Perceptual studies ventured the explanation that this was due for instance to false phonemic selection, feature unmarking, gesture reduction. VOT studies are compatible with the interpretation that +/- voice exchanges and distortions derive from discoordination between laryngeal and supralaryngeal gestures. Instead of the essentially binary relationship of VOT for voiced-voiceless pairs, speech dyspraxics show a continuous distribution of VOTs. Typically, coordination of VOT is highly variable. Sometimes this drift is not sufficient to occasion misperception of the target by the listener. Other times deviation is large enough to create the impression that a false target has been selected. However, examination of underlying VOT continua demonstrates this not to be the case.
A string of investigations has confirmed this picture - including Blumstein, Cooper, Goodglass, Statlender, Gottlieb (1980); Hoit-Dalgaard, Murry, Kopp (1983); Shewan, Leeper, Booth (1984); Ziegler and von Cramon (1986 a, b). There have been disagreements over how best to measure (inter- and intrasubject) VOT - see Walsh (1983), Tuller (1984), Ziegler (1984; 1987), Blumstein and Shinn (1984). They do not alter the basic finding that there is a group of patients who show continuity in the distribution of VOT, and groups where there appears (see below) to be distribution of values as in normals. The same picture is observed across other articulation couplings.
Itoh, Sasanuma, Ushijima (1979) made nasendoscopic recordings of velar movements in an AS and a Wernicke’s dysphasic
patient. They predicted that when subjects made a substitution perceived as oral for nasal or vice versa, if there were a linguistic, phonological disorder then they would see a pattern of movement/coupling clearly within bounds expected for perceived /n/ or /d/. If the apparent substitution was due to disrupted motoric phasing between velic state and stop release, then they expected movement towards the intended /d/ or /n/ gesture but realisation of the target too early or too late, and consequently a perceptually distorted segment produced.
For the AS patient this is what was found. Identical to the continuous range of VOT values in AS, Itoh et al found a continuum of velic occlusion times. Sometimes these were close enough to normal values not to influence listener judgements on +/- oral. Other times deviations resulted in equivocal productions or apparent substitutions when the categorical boundary was crossed. In contrast the Wernicke's subject's substitutions were associated with frank exchanges of gestures. Timings here formed two, non-overlapping, curves.
Ziegler and von Cramon (1986a) had AS and normals repeat real and nonsense words containing /V/nasal/V/ sequences. The sound pressure level (SPL) of their utterances was recorded. If the transition between oral-nasal segments is correctly phased there should be no diminution of SPL. Mis- synchronisation would be expected to produce SPL damping.
This was the finding for AS subjects. The time values for mis-synchronisation and decibel measures in SPL damping were not consistent. They spread continuously across the range of values normally linked to the discrete curves for /n/ and /d/. Sometimes an apparent frank substitution was perceived. Other times when desynchronisation span was less extensive the oral cognate of the intended nasal might be heard as an appendage - /nd; nt/ /gg; T)k/ etc. Velar mistiming thus causes both perceived substitutions and additions. The instrumental support illustrates that these are due to
motoric temporal dysfunction and not phonological mis- selection.
Other workers demonstrate equivalent phenomena with other couplings. They include Ziegler et al (1985; 1986 a,b) on lip rounding and spreading for +/- rounded vowels, where mistiming of rounding onset/offset led to misperceptions of vowels; Hardcastle, Morgan, Barry, Clark (1985), Hardcastle (1987); Edwards, Miller (1989) for evidence of general incoordination of tongue tip, blade and body; Tseng, McNeil, Adams, Weismer (1989) and Robin et al (1989) found upper and lower lip asynchrony - though the latter question whether this is different from normals, and point out that there was no systematic relationship between asynchrony and perceived inaccuracy. Tuller and Story (1987) made an acoustic analysis of co-articulatory effects of upcoming vowels on preceding fricatives. Their non-fluent (AS) group displayed extreme variability in the onset of co-articulation. Co- articulation was not markedly absent - dispelling conclusions that their fundamental phonemic or articulatory specifications were altered. Rather, the acoustic consequences of anticipation could be delayed relative to normals (measured relative to friction midpoint). Katz (1987) in a similar study, actually found conflicting data, but he used fixed 70 msec and 20 msec points from which to make his measurements. Ziegler (1989) discusses how these methodological issues might be resolved, delivering more homogeneous data.
It is not just interarticulator phasing, though, that is influenced by this putative underlying motor dysfunction. Arguably the same problem lies behind the absolute and
relative segment duration discrepancies observed in AS.
5.3.1. Segnent Duration
Ignoring for the moment what constitutes a segment and how it is extracted from the flow of speech or speech recordings, several durational features seem to be general to all AS
speakers. Despite intact range, velocity, force they are universally reported as slower than normals.
Kent et al (1983) reported lengthening of steady states and transitions. Significantly they point out that lengthening is not linear, smooth, but involves ongoing distortion, /e/ is not simply realised as /e:/ but may be a composite vowel /e E eie/. Equally // / does not become /J: / but varies in SPL and maybe articulatory posture throughout its duration. McNeil, Weismer et al (1990) found AS speakers achieved normal isometric force but maintained it unstably. Likewise, the instability in target approach, maintenance and offset correlates highly with the dysme trias of dyspraxic performance reported by McNeil, Caliguiri et al (1990).
Other studies on segment duration are not at first glance unanimous. Di Simoni and Parley (1977) found /p/ duration in their dyspraxic-dysarthric lady shorter than for normals in one to three word utterances. Duffy and Gawle's (1984) subjects seemed to produce vowels of shorter duration than normals in monosyllables. Candour and Dardarananda (1984) and Ryalls (1986) found negligible effect of AS on vowel length in monosyllables. Recourse to the fine print reveals, though, that there was great variability amongst dyspraxics' productions and data for longer utterances reveals consistent contrasts between AS and normals. Two features emerge.
Firstly mean segmental durations become longer than normals *. Di Simoni et al (1977) found this for /p/ production, Collins, Rosenbek, Wertz (1983); Ryalls (1987); Strand, McNeil (1987a) for vowels. The second feature is that linguistic and rhythmic roles of length are preserved.
Normal speakers systematically shorten stressed stem vowels in proportion to number of syllables following. Hough, Klich
(1987), Strand et al (1987a); Baum (1990) and others demonstrate that despite greater absolute length, relative length remains intact in AS. In English, in citation forms of single words, vowels preceding final obstruent consonants
differ in length according to the +/- voice properties of the final consonant. Duffy et al (1984) noted that despite absolute deviations from normals, this relative length feature remains.
It can be speculated that listeners perceive voiceless for voiced substitutions word finally because of shortened vowel length, and voiced for voiceless where vowels tend to be prolonged in longer words. Lengthening of, e.g. /I/ or /%/"/ in English leads to impressions of vowel distortion but not phonemic change to /i/ or /u/. In languages or accents where lengthening is not associated with tongue position changes, such shortening/lengthening once it reaches a critical boundary value will be heard as vowel substitution. Disordered isometric force maintenance in steady states will also lead to impressions of distortions and additions. Abnormal durations may also perceptually create the impression of aberrant stress assignment.
The fact that all the individual dyspraxics are capable of uttering the full range of vowel durations and that they keep relative durations in the face of mean absolute duration variation supports the contention that in AS there is (echoing Strand) an impaired precision mechanism in activating/executing motor plans and not a misapplied rule system.