WHAT PHONETICS HAS TO SAY ABOUT STUTTERING

CLAUDIO ZMARICHC. N.R.- Istituto di Fonetica e Dialettologia

-----Lecture held at the Socrates European Intensive Programme for Speech and Language Therapy,

Collegio Mazza, Padova, 22.08.2000 and 28.08.2000

1.1 INTRODUCTION

We begin this lecture by presenting some well known concepts and experimental facts aboutstuttering (Bloodstein, 1995):DEFINITION• “Disorder in the rhythm of speech, in which the individual knows precisely what he wishes to

say, but at the time is unable to say it because of an involuntary, repetitive prolongation orcessation of a sound” (W.H.O. 1977: 202).

SYMPTOMATOLOGY• Blocks, repetitions and prolongations of a phone are considered to be primary symptoms, and are

part of the so-called dysfluencies.EPIDEMIOLOGY• Onset: in 75% of the confirmed stutterers, stuttering begins from 18 months to 3 years, and

virtually vanishes after the 12th year.• Prevalence: 1% (in a precise moment).• Incidence: 5% (lifetime risk, for a period of at least 6 months).• Spontaneous recovery: 3 out of 4 children.As a phonetician, two main questions about stuttering are worthwhile. The first is: “Why wouldPhonetics be so important in the study of stuttering?” A possible answer could be that Phonetics is aborderline discipline, both in the sense that it has a theoretical as well as an applicativecharacter, and in the sense that it is at the convergence of different scientific realms, such aslinguistics, physics, biology, psychology etc. As a borderline discipline, it holds a privileged keyfor unifying and simplifying the understanding of stuttering that presents itself as amultidimensional phenomenon (made of sociocultural, psychological, physiological and geneticfactors). In fact, one can say that in order to accomplish a causal function in stuttering, each ofthose variables must at the end act on the motor control processes of the speech apparatus(Smith & Kelly, 1997). Phonetics has more of an advantage in the description of these speechprocesses than other linguistic disciplines because, as Lindblom (1995) said, “phonetics can invokeknowledge which is relevant to speech but which was acquired independently of it […], such asinformation on the general mechanisms of hearing and motor control”.The second question is: “Why should a phonetician be interested in stuttering?” Once again, wecoud answer that as phoneticians we feel a potential attraction towards a speech disorder thatselectively affects fluency, leaving essentially intact the syntactic and grammatical structures inindividuals that are judged to be safe and normally endowed with reference to cognitive andaffective aspects. At the same time, the speech aspects under investigation in stuttering are at theheart of a number of theories of speech production (Weismer et al., 1995). In fact, these theories canbe affiliated to two great families on the basis of their solutions to the problem of speech timing.According to Fowler (1980), theories about speech production can be divided in extrinsic andintrinsic timing theories. The first ones, also called “translation theories”, postulate the existenceof a timer, possibly not specific to the speech mechanism, which put in sequence a series of discreteand timeless units (i.e. columns of distinctive features); the second ones, also called “gesturetheories”, bring the timing organization back to the general dynamic property of the articulatory

system. The validity of these theories could - in the end - be proved by their power in explainingtiming phenomena that are characteristics of the motor speech disorders, like: articulatoryslowness, abnormal scaling (in magnitude) of the articulatory gestures, variability of speechproduction (across repetitions), segmentalization (reduction in the extent of the coarticulation).

2.1 ETIOLOGY: THE LOST PARADIGM

While past researches in the stuttering field were carried out under a general unique consensualparadigm which most researchers agreed upon, the current situation of great confusion has beenwell depicted by Siegel (1998: 108): there “now exist numerous theories [of stuttering], eachproductive in its own right, but also more restricted. […] if the further purpose of a theory is toprovide an overarching paradigm of the sort that Kuhn (1970) described, in which the participantsshare a common vocabulary, world view, and research agenda, the current theories fall far fromthe mark”. The persistent lack of a general and comprehensive explanation of stuttering, and theproliferation of experimental results pointing to different directions brought Starkweather (1987) tostate that current theories of stuttering are “descriptive”: “it turns out that the several researchessimply help to describe the disorder in a more complete way, but not to explain it”.Nonetheless, there are hopes that the situation could change positively. In the words of Siegel(1998), “although there is no dominant theory to set the research agenda, there are other sourcesthat provide some guidance”:

• Clinic: research helps to identify the conditions under which various therapy approaches leadto successful outcomes;

• Technology: new technology energizes research and stimulates theory development;• Puzzles: these can emerge when two experiments yield discrepant results.

At the same time, a common frame of reference begins to emerge, as it was summed up by Smith &Kelly (1997), when they say that the cause of stuttering cannot be understood in the traditionalsense of relationship of dependency between the explanans and the explanandum based on a logicalor a factual necessity. No single sufficient cause provokes stuttering, but rather various criticalfactors, that in some combinations can become sufficient. The cause of stuttering cannot beidentified in the experimental variable(s) common to all the stutterers and absent in all those whodon’t stutter. There is an increasing consensus in the opinion that stuttering emerges as soon as analtered relation between dynamic and multifactorial processes of a complex nature generates thosealterations of the fluency, which are judged from the common sense like anomalous.If the the cause of stuttering is presently unknown, according to Yaruss et al. (1998), that modifiedthe proposal of McClean (1990), based on the W.H.O. classification of the consequences ofdisorders, the consequences of stuttering can be described at 3 levels:.• Impairment: any loss or abnormality of psychological, physiological or anatomical structures or

functions;• Disability: any restriction or lack of ability to perform an activity in a normal manner;• Handicap: a disadvantage for an individual resulting in restrictions of social roles.

2.2. The Nijmegen conferencesIn the last two decades the view that stuttering is best understood as a neurophysiological disorderthat directly affects the speech motor system has been promoted. In particular, the University ofNijmegen (NL) organized 3 conferences, on 1985 (Peters & Hulstijn 1987), on 1990 (Peters,Hulstijn & Starkweather 1991) and on 1996 (Hulstijn, Peters & van Lieshout 1997). During the lastconference, several researches based on brain imaging technologies were presented. From thesestudies it emerges that the cerebral processes relative to speech production are stronglycompromised in stutterers and that stutterers are different from controls also at rest condition.Nearly all the studies show anomalies in the activation of the cerebral motor areas and of the

cerebellum, and all have brought back differences between stutterers and controls for cerebrallateralization.Regrettably, researches based on PET and SPECT (together with other diagnostic technologies likeEvoked Potentials, Quantitative Electroencephalography, functional Nuclear Magnetic Risonance,cf. Lauter 1997) could well highlight the brain area that are involved in speech processing but theydo not inform on the particular kind of process that is taking place. For this reason we must refer toa detailed model of the speech production, such as that elaborated by Levelt and colleagues (cf.Levelt 1989, Levelt, Roelofs & Meyers 1999). Such a model characterizes the units of processing,explicitates the mental representation of the linguistic information and puts in evidence theprocesses of elaboration of such information (selection, access, monitoring and correction etc).

Fig. 1. A blueprint of the speech production model of Levelt (adapted form Levelt, 1989)

For the part more specifically motoric, we will integrate such model with the proposals of Gracco(1997) and Saltzmann & Munhall (1989), in a way similar to the presentation of a mixed orintegrated model made by Van Lieshout (1995) and Hulstijn & van Lieshout (1997).

3.1. THE MODEL OF SPEECH PRODUCTION BY LEVELT AND COLL.

The various processing stages extending from the intention to communicate to the overt articulation

can be summed up in the following steps (fig. 1):• the speaker has a communicative intention;• he selects and orders the information necessary to express that intention in the conceptualizer;• he assigns a linguistic shape to this preverbal message, by retrieving the appropriated words

from the memory, and assigning the corrected grammatical roles and the syntactic positions tothem;

• he calculates the phonetic specification of the utterance under construction;• he uses this plan like a guide for the articulatory execution that produces overt speech.In addition, there are processes of control based on three feedbacks: the first one concurs toestimate the appropriateness of the semantic information in relation to the preverbal message, thesecond allows the inspection of the syllables of the phonological words, in the so-called “internalspeech”, and the third one is the auditory feedback of external type.The high speed of execution (10-15 phonemes articulated every second) demands that the variousprocesses operate serially but in a nearly simultaneous way (incremental production): a process ofinferior level starts as soon as the process of the closer high level has delivered a minimal part of itsproduct to it. The way to operate is highly automatic and modular. The automaticity mostlyguarantees the allocation of the attention of the speaker to the content and the modularity guaranteesthe absence of dangerous interferences among various levels.

Fig. 2. Stages in the production of words according to Levelt (adapted from Levelt, 1994)

Lexical access. The conceptual processing produces a series of lexical concepts (fig. 2). Everylexical concept activates in the mental lexicon the correspondent lemma (equipped with semantic -syntactic information), which is then transferred to the Formulator. As it happens in theconnessionistic models (Dell & Reich, 1980; Stemberger, 1985; Dell, 1986; MacKay, 1987;Roelofs, 1992) there is a network of connections among various linguistic units. In the lexicalaccess the conceptual characteristics determine the simultaneous activation, even if with various

degrees, of all the lemma semantically related. The probability of selecting a particular lemma isdetermined by the so-called Luce ratio, equivalent to the activation of the lemma target divided bythe sum of the activation of all "the closest " other lemma (Roloefs, 1992).Formulator: the grammatical encoding (Levelt & coll.) The Formulator comprises two levels ofprocessing of the utterance: the grammatical encoding and the phonological encoding. Thegrammatical encoding consists of two main processes: the functional planning and the positionalplanning (cfr. Bock & Levelt, 1994). At the functional level, the syntactic-semantic information ofthe lemmas determines a structure of the predicate/argument type, and the lemmas receive thematicroles (predicate, agent, object, beneficiary etc). At the positional level lemmas are inserted in aconstituent structure of the utterance, specifying the serial order of lemmas and affixes, thefunctional words and the stress hierarchies.Phonological encoding. The phonological encoding is the construction of a phonetic plan from thesurface structure resulting from the positional planning. Five different processes are involved: theretrieval of the lexeme, the segmental spell-out, the metrical spell-out, the metrical word formationand the segment-to-frame association:• The retrieval of a lexeme from a lemma happens with a connessionistic mechanism to which the

effect of word frequency contributes;• The spell-out of a lexeme consists at least in the segmental information (underspecified

phonemes) and in the metrical information (at least the number of syllables and the stressstructure);

• Formation of the metrical frame: union of the metrical frame of a noun with the metrical frameof its clitic: es. " demand it ” σσ’+ σ −−> σσ’σ;

• The association of the segment to its frame consists in filling from left to right the alreadyprepared metrical frame with the previously specified segments (creation of the phonologicalwords): es. /di.’m@n.dit/. The end product of the stage of phonological coding is a sequence ofsyllables, that in fluent speech can appear as fast as one every 100 ms.

• As phonological words are formed, one by one, during phonological encoding, the correspondingphonetic or articulatory syllables are retrieved from mental syllabary (see fig. 3, from Hulstijn& van Lieshout 1997).

Fig. 3. An adapted version (from Hulstijn & van Lieshout, 1997) of the Levelt model (1989), showing the processes(left part) and storage components (right part) related to speech production, starting from the phonetic coding stage.

4.1. THE PHONETIC ENCODING.

The articulatory syllable contains a specification of the articulatory task for its realization, that is agestural score (articulatory phonology, Browman & Goldstein 1986, 1997). This specificationconcerns the tasks for the individual articulatory structures of the glottis, velum and the oral cavity,and their synchronization, to accomplish constristictions of varying degrees. In this model theindividual articulatory organs are called effectors. Each effector (for instance the jaw) participatesin a coordinative structure (also called tract variable) that is different from time to time and that isa cooperative functional unit among a specific group of articulators (for ex., upper lip, lower lip,jaw). The gestures that make use of different effectors are distinguished in a categorical (i.e.“phonologic”) way. Other distinctive oppositions are created within the same coordinativestructure: distinctions among different places of constriction (velar, palatal, alveolar etc.) andamong different degrees of constriction (occlusion, friction, opening etc).The gestural score describes the degree of involvement in time of any tract variable in the gesturesrealizing the utterance, and the synchronization among them. This representation does not comecompletely specified: labial constriction, for instance, does not predict how many mm each lip hasto move. The product of the phonetic stage is a sequence of gestural scores related to the sequenceof the syllables in the phonological words, the so-called phonetic/articulatory plan. Thephonological words are grouped in the phonological syntagms (= intermediate syntagms) and thesebecome part of the intonational syntagms (cfr. Nespor,1993).Usually the process of phonetic encoding is normally in a more advanced state with reference to theslower processes relative to articulation. The part of the phonetic plan already available and waitingfor the articulatory processes is deposited in a short-term articulatory buffer, comprising at leastone phonological word (Levelt 1989, 1993) .

4.2. Phonetic encoding in stutteringThere are many studies that have found differences between the stutterers and the nonstutterers onkinematic or acoustic measures of speech, but these measures are not attributable in a direct andunivocal way to a determined stage of the model of Levelt. Of this type are the acoustic studiesthat found the speech of stutterers to be slower and more variable are of this kind, and thekinematic studies, evidencing as the articulatory gestures of the stutterers are characterized bygreater duration, minor amplitude and inferior maximum velocity than that of the nonstutterers(Zmarich, Magno Caldognetto & Vagges, 1994, 1995; for other references, cf. van Lieshout, 1995:34). What follows is a summary of the main results of the most recent researches, subdivided byLevelt’s speech processing stages attributable to the phonetic level:Substage n.1. motor plan assembly:• Phonological encoding (Postma & Kolk, 1993): Stutterers make a great amount of lapsus in the

phonetic plan (internal speech) because of delays in the phonemic selection.. They are detectedby the internal monitor and corrected, resulting in overt stuttering.

• Phonetic encoding (Peters, Hulstijn & Starkweather, 1989): the greater delays in the reactiontimes for longer words could be due to the initial (abstract) stages of motor planning.

• Creation Phonetic plan (Kent 1997, Kalveram 1997, Ludlow et al. 1997, McClean 1997): Alearning failure results in inefficient phonetic gestures.

5.1. THE ARTICULATION STAGE

The first stage is the preparation of the muscular commands, and consists in theimplementation of the gestural scores. This is accomplished in two substages necessary to convertthe phonetic/articulatory plan in specified commands to the single articulators. The first substage

(Retrieval and unpacking of motor plan) regards the retrieval of the motor routines (in thephonetic/articulatory plan) from the short term motor buffer and their subsequent decomposition incommands to the individual muscle (cf. Sternberg et al., 1978). The second substage is the muscleforce adjustments, and regards the parametrization of the single muscular commands. The thirdand last stage consists in the execution of the gestural scores. The gestural scores do not specifythe motor programs completely. Because of the possibility of a temporal overlapping betweenadjacent gestures, the movements of the articulators will not show contextual invariance, due to thefact that some articulators can be part of the coordinative structure of another gesture that is beingproduced at the same moment. The dynamic specification contains the parametrization values of adynamic system regulating the formation of constrictions and releases from the part of the pertinenteffector. The dynamics specification comprises: the equilibrium point (the target of the tract-variable), stiffness (connected to the temporal constant of realization of the constriction) anddamping. The gestural scores are interpreted by the Task Dynamic Model (Saltzman & Munhall,1989), which calculates the answer (variable in time) of a group of articulators of the vocal tract tothe imposition of that particular dynamic system, the transfer function of the acoustic energy andthe finally produced wave of acoustic pressure.

Fig. 4. A schematic model of hypothetical speech motor process. The thickness of the lines reflects the strenght of theconnections; double headed arrows represent interactive connections (from Gracco, 1997).

The Task Dynamic Model could be considered as an introduction to the neurobiological model offluent speech production put forward by Gracco (1997). The end-product of the processes of thephonological encoding is a sequence of syllables that are already equipped with an intrinsicrhythmic organization. The phonetic content of each syllable regulates the instantaneous frequencyof the articulation but does not modify the general rhythm.The characteristic frequencies of thearticulator can easily be modified by simple inputs of a tonic nature which are part of thepreparatory regulation.The intention to speak determines a general increase of the levels of arousalassociated to the sensory and motor regions, and the specific regions involved in speech productionare tuned to the upcoming motor commands. The overall levels of excitability are dependent onsome variables of pragmatic type (emotional content, speaking rate, environmental noise etc.),which determine the total level of effort.Gracco (1997) states that “the syllabic configurations stored in the syllabary, together with the

neural commands which specify the coordination between the gestural constituents of theconfigurations, converge on and interact with the preparatory activations on output areas. […] Thecentral commands that give rise to specific and neuroanatomically focused motor patterns result in amovement-to-movement tuning of the motor output structures in characteristic ways consistent withthe articulatory organization for unique vocal tract configurations. […] The motor state space is alsoconditioned by ongoing topographic projections from somatic sensory vocal tract representationswhich have to been made active by the preparatory process”.The articulatory feedback. The speaker continuosly modifies the execution of the complexmovements for speech production in answer to the changes of the organic or environmentalconditions. This suggests the existence of a finely tuned ability to feel variations of the conditionsof the vocal tract, and to incorporate those changes in the planning and execution of the speechproduction processes. There is a theoretical and experimental lack on this topic, in some measureattributable to the infuence of a type of model based on preprogramming (MacNeilage, 1970),which did not leave any space to the influence of the proprioceptive feedback, considered to be amodality of afference which is too slow to be in a position to modify the articulatory behaviors online. More recent experiments on the role of the propriocetptive feedback revealed that in case ofimmobilization or anesthetization of an articulator (for instance the jaw), the other articulatorswithin the same coordinative structure (for instance the lips) are able to compensate in lessthan 50 ms the lacking contribution, thus permitting the attainment of the acoustic target.The interaction of sensory-motor information with neuromotor patterns generated at the centrallevel reduces the computational load caused by the contextual variations. Kinesthesis is importantduring the execution of the multiarticulated, expert, and precise movements, because it supplies theSNC with the necessary flexibility to control such sequences. The kinesthetic signals constitute animportant part of motor learning and contribute to the continuous updating of the motor controlstructures by regulating the movements presently under execution and the to-be-executedmovements (Schmidt, 1988).

5.2. The articulatory feedback in stutterersGenerally speaking, the kind of feedback considered to be more relevant to stuttering is theproprioceptive or kinesthetic feedback, but this topic has not received much attention.Someone claimed that stuttering involves a problem in the auditory feedback (cf. referencesquoted by van Lieshout 1995). De Nil & Abbs (1991), De Nil (1995) and Archibald & De Nil(1999) found stutterers to be less sensitive to detect the sensory information relative to the oralstructures. Their threshold level was too high, therefore produces minimal movements alwaysgreater than those produced by controls.These results suggest that stutterers could have problems in the evaluation and the exploitment ofthe sensorimotor information.Muscle command preparation stage:• Muscle force adjustments. (De Nil 1995, Archibald & De Nil 1999): the sensorial information

of proprioceptive kind on the position and velocity of the articulators is defective (highertresholds in propriocepsis);

• (Grosjean et al. 1997, Howell et al. 1997, Zelaznik et al. 1998): the control of the muscularparameters is defective (the force is reduced and more variable);

• (Van Lieshout 1995): bad trade-off between speed and accuracy in performing a speech task(Fitts’ law);

• Initiation: timing of the neural discharge (McClean, 1997): there is a distinct neural loop forinitiation, and the triggering function associated with these recurrent loops may be impaired inspeech disfluency;

• property of the neural discharge (Starkweather, 1995): ineffective force discharge (excessiveEMG activity).

• Initiation: coordination see later

Muscle command execution stageGeneral considerations: the same factors preventing a correct planning of the motor commandbefore the beginning of the utterance could be present during the motor planning of the last units ofthe utterance, simultaneously to the articulation of the first sounds. The sensory afference isnecessary not only to the selection and regulation of the parameters before that movements takeplace, but even during the same movement.• Muscle command execution: (many authors, cf. references in Croatto & Zmarich 1992,

Zmarich et al. 1994, Zmarich 1998, van Lieshout 1995): a deficitary or unefficient executiongenerates kinematic anomalies;

• defective proprioceptive and kinesthetic feedback processing (De Nil 1995, Neilson &Neilson 1987);

• defective auditory feedback processing (Howell 1987, Kalinowsky et al. 1995);• learning deficiency in “audio-phonatoric coupling” (Kalveram & Natke, 1997).

6.1 DEFECTS OF THE TRADITIONAL ATTITUDE TOWARDS THE STUDY OF THEPATHOLOGICAL SPEECH

Traditionally, the description, explanation and therapy of several speech disorders have beenbased on the phonetic transcription (from the auditory perception) and on the electroacousticanalysis. These methods present restrictions for each of the above-mentioned three levels.The restriction in the description is due to the subjective character of the assesment of theauditory percept, and in the absence of biuniqueness between the acoustic and the articulatory data.With reference to the phonetic classification of the auditory percept, a study of Shriberg & Lof(1991) on the reliability of the phonetic transcription emphasized that the phonetic trascription ofthe “narrow” type presents percentage scores of inter-iudges agreement of 74%, which arereduced to 34% when diacritics were taken into consideration. The authors concluded with aninvitation to use technological devices for the acoustic analysis.But the acoustic data could be ambiguous too, due to the absence of both a common standard forthe acoustic cues and the analysis procedures for identifying the phonetically relevant parameters,and the absence of biuniqueness between the acoustic datum and the articulatory datum.Nontheless, the traditional attitude in the explanation is also constrained by the opacityintroduced by the relative distance between the more or less central cause of the pathology and themore or less distal periphery in which the acoustic or perceptive events are measured and theinadequacy of the phonetic-phonological theories based on perceptive or acoustic targets to explainmotor events of intrinsically dynamic nature.The dynamic models of the articulatory production better explains the characteristic phenomena ofspeech disorders of motor type affecting phonetic gestures, such as:• articulatory slowing down, even under instruction to speak at a rapid rate (reason: the

temporospatial specification of a gesture is inherently deviant);• abnormal scaling (the regulation of magnitudes of articulatory displacement as a function of

time). Error of scaling are a means of explaining hypokinetic and hyperkinetic articulatorymovements;

• Incorrect Phasing of Gesture: produce a “sliding” of gestures that potentially alters the phoneticpercept;

• excessive variability: unreliable timing of speech movements;• segmentalization (reduction in coarticulation): separation of the normal overlapped pattern of

speech movements;• restricted adaptability to scaling variables: a reduction in the ability to modify the overall

speech pattern in the presence of prosodic and stylystic variations;• deterioration of the Gestural Score.

There also exist constraints in the way treatment is traditionally considered and implemented, due tothe inability for the therapy to give adequate feedback of articulatory nature to the patient.Stuttering has an intermittent and multidimensional nature, and can be described at multiple levels:sociocultural, psychological, physiological and genetic.The traditional definition and diagnosis ofstuttering are based on the auditory detection and qualitative assessment of the dysfluencies(abnormal for number, type, duration and position). The traditional attitude identifies the lociof the stuttered utterance that are associated with the occurrence of the disfluencies and explainsthese distributional patterns by invoking the same dysfunctions of the mental representations andprocesses which generate lapsus and disfluencies in normal speakers (Zmarich 1991). Its maindefect rests on the exclusive attention to the disfluencies, which render the speech “discontinuous".But fluency is multidimensional: not only is the fluent speech devoid of discontinuities, but it isalso produced with a regular rhythmic beat, in rapid rate and without excessive physical and mentaleffort (Starkweather, 1987). There are stutterers without disfluencies: they are affected by"covert/subperceptual stuttering " and often perceive in speaking excessive levels of muscular effortand “cognitive tension " that can pass unobserved to the eye and the ear of the clinician.

7.1. A NEW WINDOW ON STUTTERING

Description. Kinematics (which describes the movement in terms of duration, extension and all oftheir derivatives such as velocity and acceleration) and dynamics (which describes the physicalconditions responsible for every given movement, and which encloses the coefficients of mass,rigidity, damping etc; cf. Bingham, 1988) always offer a useful integration to the traditionalmethods of description, by being able to give a qualitative and quantitative analysis, reliable andexhaustive, of the movements of the articulatory organs (Gracco, 1992).For instance, the kinematic description produces a more accurate and individualized description ofthe peripheral physiology of disfluencies (Zimmermann, 1980). Together with electromyographicdata, kinematics and dynamics deepened our knowledge on the anomalies of muscular contractionin stutterers (Fig. 5).

Fig. 5. Lower lip movement and EMG tracings of the Mentalis muscle and the Depressor Labii Inferior muscle indisfluent pronounciation of / p / for subjects 2-5 (soggetto 2,3,4,5). From the top down in each figure the records arelower-lip vertical displacements. For subject 2 and 3 the arrows indicate the onset of lower-lip depression forproduction of the intended vowel. For subject 5 the fourth record down is the audio signal (from McClean et al. 1984,modified)

Explanation. For their nature, kinematic and dynamic data lend themselves very well to beenclosed in an analogous theoretical frame, which is for instance the non linear dynamic theory(cfr. Port & van Gelder, 1995), that quite recently has been applied to the field of the motor control(cfr. Kelso, 1995). According to this theory, the main unit of the speech motor control (and theobject of the motor learning) is the so-called phonetic gesture (cf. the recent reviews in Browman &Goldstein, 1997; Saltzman,1999). Basically, the adoption of a dynamic theoretical frame permits agreater facility in the formulation of etiologic hypotheses based on speech motor control. Forinstance, an important distinction between normal and abnormal speech is that the movementsrelative to the first one has a voluntary nature, while the disfluencies of the stutterers do not.Another advantage is the possibility to use reductionism in a etiological key: “entia non suntmultiplicanda praeter necessitatem”. As to stuttering, the alleged variables of sociocultural,psychological, physiological and genetic nature play a role only to the extent they influence theprocesses of the motor control of the speech apparatus, which are the direct cause of the stutteringepisodes. Further, this approach favours the possibility of investigating the speech production of thestutterers which is perceptively fluent, in order to establish if they are pathological subjects alsowhen speaking they do not produce disfluencies. In fact it can be useful to look for the anomalies inthe perceptively fluent speech of stutterers on the basis of the claim that they would be in some waycloser to the nature and the origin of the problem (according to non linear dynamics, minimalvariations can carry to vaste structural changes). In this way, every reference to reactions (=symptoms) of secondary and learned nature would be eliminated.The anomalies would have to be identified by comparing the kinematic patterns exhibited by thestutterers with those exhibited by the nonstutterers. However, some notes of caution are due, owingto the fact that the ability to generate an invariant acoustic product through a (restricted) series ofdifferent configurations of the vocal tract is a characteristic of the fluent speech production(adaptation to continuously changing phonetic contexts). According to Folkins (1991) two types ofadaptation of the articulatory system exist, which are conservative of the phonological function:• Flexibility, which is responsible of the articulatory modifications due to segmental and

suprasegmental coarticulation;• Plasticity, which is responsible of the general equilibrium in which these modifications take

place, and eventually allows fast transitions towards new states of equilibrium (for inst., speakingat the same time whilst eating or smoking).

It is possible to consider the physiology as pathological only if it overcomes the “normal” limit offlexibility and plasticity.Figure 6 shows the displacements of the closing gestures in a sequence of / pa / syllables. Althoughthey had been realized by the stutterer (left side) with extension and velocity values that were verydifferent from those produced by a normal subject (articulating with the same preferentialfrequency, right side), they turned out to be perceptively fluent (Zmarich et al.1995).

Fig.6. Plot of individual peak velocity and displacement values as a function of four different speaking rates (ms):100=50-100; 150=101-150; 200=151-200; 250=201-250. The articulators are: U=Upper Lip, L=Lower Lip, J= Jaw. Thestuttering subject is on the left side, the nonstuttering subject on the right (adapted from Zmarich et al., 1995).

50100

150200

250300

0

510

15

20

100

200

L

J

L

J

UL

J

U L

J

DURAT (msec)

DISPL (mm)

VEL (mm/sec)

50100

150200

250300

0

510

15

20

100

200

U

LJULJ

ULJ

UL

J

DURAT (msec)

DISPL (mm)

VEL (mm/sec)

The fact that fluent stutterers present very different articulatory configurations from those ofnonstutterers does not imply by itself that these are manifestations of stuttering, but it couldimply a scarce, but not deficitary, ability, or, alternatively, it could reflect the exploitation of anhealthy part of the speech system in order to react to the part affected by stutteringIf we want to use kinematics with etiological function, we must therefore keep in mind theexistence of a great intra- and inter-individual variability in normal speakers. From hereon thereis a need to study the articulatory movements of the stutterers fluent speech on a individual basisand on a great number of repetitions (if the anomalies are involuntary it is probable that they willincrease the variance of the kinematic patterns). At this point there is also the need also to excludethe kinematic descriptions - that use duration, extension and velocity directly - from the comparisonbetween the movements of stutterers and nonstutterers, in order to base our assesment only onthose dynamic properties that remain relatively invariant (stable) through more repetitions.Here below are indicated some indexes of articulatory stability.Among the indexes for the individual articulator there is the presence of a unique peak in thevelocity trace: figure 7 illustrates the curves of the instantaneous velocity of the articulators of astutterer engaged in repeated gestures of bilabial closing and opening, at comfortable rate (Zmarich& Magno, 1997).

Fig. 7. Displacement and velocity profiles vs. time for each articulator, here represented together with the acousticsignal. Critical analysis points are also highlighted in the UL velocity profile (from Zmarich et al. 1997).

These curves are normally characterized by a single peak of velocity for each gesture, butsometimes, it happens, as in this case to the upper lip (UL), that some extra-peaks can be noticed.In this experiment, the velocity curves of the gestures realized by the stutterers presented a numberfrom 3 to 5 times greater than controls (according to the articulators) of multipeaked curves.These result has been explained with the hypothesis that stutterers make a more intense andcontinuous use of the mechanism of proprioceptive feedback (considering that, according toAdams, Weismer & Kent 1993, the presence of multiple peaks reflects the employment of a number

of submovements having a function of temporal and spatial regulation during the main movement).Among the indices for inter-articulatory (paradigmatic) coordination there is the profile of thevelocity curve of the coordinative structure (Caruso, Abbs & Gracco, 1988; Alfonso, 1991): itresults from the weighted sum, moment by moment, of the movements of the individualarticulators. For instance, if the target is constituted by a bilabial plosive, the coordinative structureis formed by the lips and the jaw. In fact, individual articulators, in fact, can contribute in a variableway to the realization of the target, but this is relatively invariant and characterized by a smoothand single-peaked velocity curve, on which more repetitions would come to coincide with theirtracings. In the experiment of Zmarich & Magno (1997) the stutterers produced velocity curves ofthe coordinative structures with multiple peaks 5 times more than nonstutterers did (8.9% vs.1.6%).An index for the inter-articulatory (paradigmatic) coordination that has already demonstrated to beparticularly useful is the motor equivalence covariability (Alfonso, 1991): it reflects the synergicway with which the coordinative structure is organized in order to realize an articulatory target.The covariability is measured through the coefficient of variation (CV) on many repetitions. Suchcoefficient, expressed as the arithmetical ratio between the standard deviation and the mean,standardizes the variability of the movements of the individual articulators. Lower scoresreflect a greater stability, normally associated to the parameter of the coordinative structure (i.e. theweighted sum of displacements of individual articulatory movements).Figure 8 shows the CV of the displacements of the articulators involved in the bilabial closinggestures for the initial consonant of / pap /, produced by stutterers (Alfonso & van Lieshout, 1997).The bars with the arrows show that the CV of the combined articulators (= coordinative structure)are greater than those of the upper lip.

Fig. 8. Coefficient of variation for the /p/ closure gesture of stuttering subjects. Synergies that do not meet motorequivalence covariability are indicated by arrows (from Alfonso & van Lieshout, 1997)

We consider now the indices of articulatory stability for the inter-gestural (syntagmatic)coordination: under this label we find the indices of coarticulation, which measure, typically, thearticulatory modifications of a phone’s speech gesture because of the influence of adjacent phones(Farnetani, 1995).Alfonso, Watson & Baer, (1987) measured the closing gesture for the first / t / of /e' tete/ producedin a perceptively fluent manner by a nonstutterer and a stutterer. The stutterer realized theocclusion with an excessive elevation of the jaw, which consequently forced the dorsum of thetongue to lower in order to be able able to coarticulate the subsequent vowel.Following this experiment, Alfonso (1991) implemented the configuration produced by thestutterer in an articulatory synthesizer (Rubin et al. 1981) and obtained the requested acoustic

target, but as soon as he tried to simulate a further lowering of the jaw, the acoustic productioncame to a stop, thus demonstrating, indirectly, the value of critical threshold of that configuration(Fig.8). This experiment clearly illustrates the value of the virtual simulations of human behavior.In an articulatory gesture one should always make a distinction between the stability (positiveconnotation) from the rigidity (negative connotation), and the flexibility (positive) from theunpredictability (negative). Stability and flexibility are the two sides of the same coin: adynamic system is stable when it allows a (limited) number of alternatives for the attainment of atarget. These alternatives are controlled variations and can have different energetic costs that willdetermine their use. A way to estimate stability consists in pushing the articulatory system closeto or also beyond its own limits. The simpler way is to individuate the highest articulation rateperformable without alterations of the auditory percept. The rationale behind this strategy is theknowledge that the time of execution of a movement is determined by the length of the trajectoryand the precision required in the attainment of a target (Fitts’ law, 1954). When the stutterers areforced to be fast, they cannot be precise anymore. This behavior is consistent with the hypothesisthat stutterers have a smaller amount of “resources” at global level.

Fig. 9. Jaw and tonguetip trajectories for various modifications of the articulator weights associated with jawmovements. Thick lines represent trajectories generated in the normal condition. Dashed and thin lines representincreasing jaw contribution to alveolar closure and release, and the compensatory action of tongue displacement. Seetext for further details (caption and figure from Alfonso, 1991).

The relative continuous phase index is a more elaborated version of this procedure which at thesame time uses the movement and the velocity of two articulators pertaining to the samecoordinative structure (van Lieshout et al. 1995). It becomes therefore possible to find theorganization of the alternative coordinated patterns, which typically passes from the structuredalternatives (at the lowest rates), to the coordinative, more stable, pattern (at maximal rate), andfinally to the breakdown of the coordination (and articulation) when the requested rate is notlonger sustainable by the stutterer.Among other indexes, the active manipulations of the articulatory behaviors are believed to be“the most powerful experimental” paradigms by Folkins (1991: 567), because they “manipulatephysiological parameters so that functional linkages can be examined directly rather than inferredfrom correlations obtained during passive variations”. A research strategy for estimating the

flexibility (the funtional interdependence) is the one used to generate interferences. It can beperformed, for instance, by suddenly blocking an articulator during the production of a phoneticgesture or by applying a force of contrary sign to it and observing the effects on the articulatorycoordination of the other articulators involved. In the same vein, it has been demonstrated (Bauer etal. 1995, 1997) that the effects of the mechanical perturbation of the movement of the jaw duringthe opening of the alveolar constriction for / s / in / sasasa / are much more destabilizing for thestutterers who had previously been classified as severe.The most recently proposed and very interesting index is the spatiotemporal index, or STI (Smithet al.1995; Smith, 1997). It derives from the idea that the articulatory production of the normalspeakers is a motor behavior highly practiced and therefore extremely stable. If we apply atemporal and spatial normalization to the kinematic tracings of the articulatory gestures relativeto many repetitions of the same utterance, they will tend to coincide in a single unique trace. Theindex measures this tendency: the greater its value, the stronger the deviation from a given trace,and therefore the greater the instability of the articulatory system. When this index wascalculated for the articulatory behavior of the stutterers, in relation to variations of the rate ofarticulation, it turned out to be much more unstable than that one of the controls.

8.1. THERAPY: TECHNICAL DEVICES FOR THE REDUCTION AND MANAGEMENTOF STUTTERING

The last two topics of the lecture will discuss considerations which hopefully are closer to theprofessional interest of speech therapist but at the same time sufficiently related to to the previouspresentation., as regards subject and point of view. I begin with a presentation of some technicaldevices for the reduction and management of stuttering. Bakker (1999:271) lists several existing“systems for assisting and automating feedback on stuttering severity, establishing speech changesthat enhance fluency, and modifying speech-related sensory feedback for fluency enhancement”.The objectives of different approaches, important to the effective clinical management of stutteringproblem, are the following:

• Stuttering prevention;• Stuttering modification or “stutter easy” approaches;• Reduction of stuttering and its related clinical behaviors;• Fluency enhancement through changing aspects of speech production (acoustic or

physiological);• Fluency enhancement through changes in speech-related sensory feedback;• Reduction of fears, concerns and negative attitudes.

Bakker focused only on the last four objectives, since the first two still rely heavily on theperceptual skills of clinician and clients, they are not supported, for the time being, by anytechnological devices. The reader interested on receiving more information about the technologicaldevices mentioned here below is invited to make reference to the article of Bakker (1999). Theeffectiveness of any feedback related effects is stated by Bakker (1999) to be greater when it isprovided immediately after the targeted behaviors

Reduction of stutteringThis objective is implemented by lowering the frequency of disfluencies and other associatedclinical signs and sometimes by reducing their duration. The main road to it resides in the provisionof performance-related feedback to clients in therapy.Technological tools. The current and general status of speaking behaviors that are often targets oftreatment (e.g. percentage of syllables stuttered, speech rate, and speech naturalness), are theavailable output of the Stuttering Treatment Rating Recorder (STRR, Fowler & Ingham), theComputerized Scoring of Stuttering Severity (CSSS, Bakker & Riley) and the Stuttering

Measurement Module (Bakker, Ingham, Ingham-Costello, Frank & Fingland). This information isperiodically given to the client through the feedback screen of the computer.

Modifications of speech production patternsMost clients can achieve immediate and substantial reductions in severity of stuttering by changingtheir speaking production patterns through the use of: the following techniques:Rhythmic production: can be established at any selected rate by using hardware and softwaremetronomes. The client has to pronounce one syllable in time with each beat produced by, forexample, Kay Elemetrics Facilitator (Kay Elemetrics Corp.), Pace Master (Associated HearingInstruments).Rate modification: is based on the count of the syllables made by the clinician using one of thesoftware device such as the above mentioned STRR or the Speech Rate Monitor and displayed in anumeric or graphical format. Another way to facilitate rate modification is through an oral-readingoriented metronome. These metronomes display the linguistic units to be produced at the pre-selected rate and clients have to read them as soon as they appear or are highlighted on the monitor.Even the use of the delayed auditory feedback (DAF) allows a reduction of the speech rate, but itsuse is complicated by the insensitivity of some client to this procedure and by a lack of precisionwhen trying to attain the planned durations precisely.Easy Onsets: according to Bakker (1999: 274), “objective feedback from instrumental measures ofvoice onset abruptness is not a standardized technology [...]”. However, the use of laryngeal,contact microphone that respond to voice-related vibrations in the soft tissues of the neck is a quitereliable method, although it remains to be determined whether the fluency enhancy effect isattributable to the phonatory process or to the articulatory and resonance systems of the vocal tract,or both. Also available are Voice Onset Monitor and the hardware/software device Dr. Fluency,“which is able to provide multimedia feedback of both acoustic (i.e. voice-onset abruptness andsegmental duration) and physiological (i.e. speech breathing) measures” (Bakker 1990: 275).Another product is the Computer Aided Fluency Establishment Training (CAFET), which providesfeedback on respiratory, voice onset, and continuity skills believed to be related to fluency.Feedback of continued phonation and airflow: There are not many systems that provide feedbackon ongoing phonation and airflow. “A system that provides visual feedback of the relative durationof phonation intervals in speech” is currently in development (Ingham, 1999). “Indirect feedback onthe continuity of both phonation and airflow may be obtained with the Cafet system”.

Fluency enhancement through altered feedbackDelayed Auditory Feedback: Besides using DAF as a methodology for establishing designatedspeaking rates, DAF is mainly used for its capacity to produce immediate and substantial reductionsin speech dysfluency for many stutterers. Among the several systems providing DAF effect, theDesktop Fluency System and the Poket Fluency System are relatively inexpensive.Frequency Altered Feedback (FAF): by increasing or reducing the fundamental frequency ofone’s speaking voice the personal quality of speech is altered substantially. This effect has beenfound to work as a fluency enhancing effect for many, but not for all, stutterers. A commercialdevice is for ex. the Fluency Master.Amplification of Fundamental Frequency: it amplifies the speaker’s voice through recording ofvocal fold oscillation with a contact microphone (the Fluency Master).Speech Masking: the Edinburgh Masker was the first technology using “bilateral masking noise,which was triggered by a voice-actuated switch to prevent unnecessary interference withcommunication [...] Voice-actuated masking is also a standard feature of the Facilitator and theDesktop Fluency System [...]” Bakker (1990: 277).

9.1 TRADE-OFF BETWEEN THE CLINICS AND THE RESEARCH

We would like to conclude this presentation turning back to Siegel’s statement about one of thethree “guidance” to the discover of the nature of stuttering: clinical research helps to identify theconditions under which various therapy approaches lead to successful outcomes.Alfonso et al. (1991) compared respiratory, laryngeal and supralaryngeal kinematics of stutterersimmediately before and after successful completion of two intensive therapy programs, only one ofwhich emphasized highly structured and physiologically oriented clinical targets. It was assumed, infact, that the most important kinematic modifications leading to fluency will be shared by allsuccesful stutterers in their post-therapy perceptually fluent speech, indipendent form clinicalinstruction. The results demonstrated that post-therapy improvement in fluency co-occurs withspatial and temporal adjustments in each of the respiratory, laryngeal, and supralaryngeal systems.In particular, as Story et al. (1996) demonstrated on the subjects treated with a fluency shapingprogram, they increased the acoustic durations, decreased the temporal and spatial extension of thespeech movements, and increased inspiratory volume and average expiratory flow.McClean, Lewandoski & Cord (1994) confirmed the increased duration of the speech movementsof treated stutterers, but the more disfluent subjects tended to show reduced variability in timingdurations of their fluent speech. This effect was attributed to volitional control intended to facilitatespeech fluency.More recently, Stager et al. (1997) compared the effects of fluency-evoking conditions (choralreading, DAF, Metronome, Noise) on aerodynamic variables to determine if any amplitude ortiming changes in flow and oral pressure were related to improved fluency. Larger improvements influency occurred under conditions when peak flow and peak pressure values were decreased frombaseline. The variables that were modified by both groups (nonstutterers and stutterers) whenspeaking under the above conditions were the same variables related to changes in fluency for thepersons who stutter.Although these results are in contrast with that of Story et al. (1996) with reference,to the findingfor the expiratory flow, they can be explained by the fact that the subjects of Story et al. probablyapplied the therapeutic instructions to reduce and slow articulatory movements and for speakingwith a breath-voice quality.

10. CONCLUSIONS

Kinematics and dynamics are useful to the speech pathologist for the description, explanationand treatment of stuttering, because they allow them to use indices which are not trivial andwhich reflect the concepts of stability and articulatory flexibility.

REFERENCES

Adams, S.G., Weismer, G. & Kent, R.D. (1993), Speaking rate and speech movement velocityprofiles, J. Speech Hearing Res., 36, 41-54.

Alfonso, P.J. (1991), Implications of the concepts underlying task-dynamic modeling on kinematicstudies of stuttering, in H.F.M., Peters, W., Hulstijn & C.W., Starkweather (op. cit.), 79-100.

Alfonso, P.J., Watson, B.C. & Baer, T. (1987), Measuring stutterers' dynamical vocal tractcharacteristics by X-ray microbeam pellet tracking, in H.F.M., Peters & W., Hulstijn (op. cit.),141-150.

Alfonso, P.J. & van Lieshout, P.H.H.M. (1997), Spatial and temporal stability in obstruent gesturalspecification by stutterers and controls: Comparison across sessions, in W. Hulstijn, H.F.M.Peters & P.H.H.M. van Lieshout, (op. cit.), 151-160.

Archibald, L. & De Nil, L.F. (1999), The relationship between stuttering severity and kinestheticacuity for jaw movements in adults who stutter, J. Fluency Disorders, 24, 25-42.

Armson, J. & Kalinowski, J. (1994), Interpreting results of the fluent speech paradigm in stutteringresearch: difficulties in separating cause from effect, J. Speech Hearing Research, 37, 69-82.

Bingham, G.P. (1988), A note on Dynamics and Kinematics, Haskins Laboratories SR-93/94, 247-251.

Bakker K. (1999), Clinical technologies for the reduction of stuttering and enhancement of speechfluency, Seminars in Speech and Language, 20, 271-280.

Bauer, A., Jancke, L. & Kalveram, K. (1995), Mechanical perturbation of the jaw during stutterers’and nonstutterers’ fluent speech, in C.W., Starkweather & H.F.M., Peters (Eds.), Proceeding ofthe First World Congress on Fluency Disorders, Munich, Germany, August 8-11, 1994, 31-34.

Bauer, A., Jancke, L., Kalveram, K. (1997), Mechanical perturbation of the jaw during speech instutterers and nonstutterers, in W., Hulstijn, H.F.M., Peters & P.H.H.M., van Lieshout (op. cit.),191-196.

Bloodstein, O. (1995), A handbook on Stuttering, National Easter Seal Society, Chicago.Bock J.K. & Levelt W. (1994), Language production. Grammatical encoding, in M.A. Gernsbacher

(Ed.), Handbook of Psycholinguistics, Academic Press, S. Diego, 945-954.Browman, C.P. & Goldstein, L. (1986), Towards an articulatory phonology, Phonology Yearbook,

3, 219.Browman, C.P. & Goldstein, L. (1997), The gestural phonology model, in W., Hulstijn, H.F.M.,

Peters & P.H.H.M., van Lieshout, (op. cit.), 57-72.Caruso A.J., Abbs J.H. & Gracco V.L. (1988), Kinematic analysis of multiple movement

coordination during speech in stutterers, Brain, 11, 439-455.Croatto, L. & Zmarich, C. (1992), La balbuzie. Relazioni tra fattori linguistici e motorii, Acta

Phoniatrica Latina, XIV, 275-299.Daniloff, R.G. & Hammarberg, R. (1973), On defining coarticulation, J. Phonetics, 1, 239-248.Dell, G.S. (1986), A spreading activation theory of retrieval in sentence production, Psychological

Review, 93, 283-321.Dell, G.S. & Reich P.A. (1980), Toward a unified model of slips of the tongue, in V.A. Fromkin

(Ed.), Errors in linguistics performance, Academic Press, New York, 273.De Nil L. F. (1995), The role of oral sensory feedback in the coordination of articulatory

movements in adults who stutter: A hypothesis, in C.W. Starkweather & H.F.M. Peters (eds.),Proceeding of the First World Congress on Fluency Disorders, Munich, Germany, August 8-11,1994, 19-22.

Fitts, P.M. (1954), The information capacity of the human motor system in controlling theamplitude of movement, J. Experimental Psychology, 47, 381-391.

Folkins, J.W. (1991), Stuttering from a speech motor control perspective, in H.F.M., Peters, W.,Hulstijn & C.W., Starkweather (op. cit.), 561-570.

Fowler, C.A. (1980), Coarticulation and theories of extrinsic timing”, J. Phonetics, 8, 113-133.Gracco, V.L. (1992), Analysis of speech movements: practical considerations and clinical

application, Haskins Laboratories SR-109/110.Gracco, V.L. (1995), “Central and peripheral components in the control of speech movements”, in

F., Bell-Berti & L.J., Raphael (eds.), Producing speech: Contemporary issues (for KatherineSafford Harris), New York, AIP Press, 417-432.

Gracco V.L. (1997), A neuromotor perspective on speech production, in W. Hulstijn, H.F.M. Peters& P.H.H.M. van Lieshout (Eds.), Speech production: Motor control, brain research and fluencydisorders, Excerpta Medica, Amsterdam, 37-56.

Grosjean M., van Galen G.P., de Jong P., van Lieshout P.H.H.M. & Hulstijn W. (1997), Isstuttering caused by failing neuromuscular force control ?, in W. Hulstijn, H.F.M. Peters &P.H.H.M. van Lieshout (Eds.), (op. cit.), 197-204.

Howell, P., El-Yaniv, N., Powell, D.J., Factors affecting fluency in stutterers when speaking underaltered auditory feedback, in H.F.M. Peters & W. Hulstijn, (op. cit.), 361-369.

Howell, P., Au-Yeung J. & Rustin L. (1997), Clock and motor variances in lip-tracking: acomparison between children who stutter and those who do not”, in W. Hulstijn, H.F.M. Peters& P.H.H.M. van Lieshout (Eds.), Speech production: Motor control, brain research and fluencydisorders, Excerpta Medica, Amsterdam, 573-578.

Hulstijn, W. & van Lieshout, P.H.H.M. (1997), Speech motor production and dysfluency, NICIReport, 97-02.

Hulstijn, W., Peters, H.F.M. & van Lieshout, P.H.H.M. (1997), Speech production: Motor control,brain research and fluency disorders, Excerpta Medica, Amsterdam.

Kalinowski, J., Armson, J., Stuart, A., Hargrave, S., Sark, S., Macleod, J. (1995), Effect ofalterations in auditory feedback on stuttering frequency during fast and normal speech rates, inC.W., Starkweather & H.F.M. Peters (Eds.), Stuttering: Proceedings of the first world congresson fluency disorders, The International Fluency Association, University Press Nijmegen, TheNetherlands, 51-55.

Kalveram, K.T. & Natke U. (1997), Stuttering and misguided learning of articulation andphonation, or why it is extremely difficult to measure the physical properties of limbs, in W.Hulstijn, H.F.M. Peters & P.H.H.M. van Lieshout (Eds.), (op. cit.), 89-98.

Kelso, J.A. (1995), Dynamic Patterns, Cambridge MA, MIT Press.Kent, R.D. (1997), Speech motor models and developments in neurophysiological science: New

perspectives, in W. Hulstijn, H.F.M. Peters & P.H.H.M. van Lieshout (Eds.), (op. cit.), 13-36.Kuhn, T.S. (1970), The structure of scientific revolutions (2nd ed.), University of Chicago Press,

Chicago.Lauter, J.L. (1997), Noninvasive brain imaging in speech motor control and stuttering: choices and

challenges, in W. Hulstijn, H.F.M. Peters & P.H.H.M. van Lieshout (Eds.), (op. cit.), 233-258.Levelt, W.J.M. (1989), Speaking: from intention to articulation. Cambridge, MA, Mit Press.Levelt, W.J.M. (1993), Language use in normal speakers and its disorders, in G. Blanken, J.

Dittman, Grimm H., J.C. Marshall & C-W. Wallesch (eds.), Linguistic disorders andpathologies, W. de Gruyter, Berlin, 1-15.

Levelt, W.J.M. (1994), On the skill of speaking: How do we access words ?, in Proceedings of theInternational Conference on Spoken Language Processing, Yokohama, Japan, vol. 4., 2253-2258.

Levelt, W.J.M., Roelofs, A. & Mejer, A. S., A theory of lexical access in speech production,Behavioral and Brain Sciences, 22, 1999, 1-75.

Lindblom, B. (1995), A view of the future of phonetics, in K., Elenius & P., Branderud (Eds.),Proceedings of the XIIIth International Congress of Phonetic Sciences, vol. 4, Stockholm, 462-9.

Ludlow, C. L., Siren K. & Zikria M. (1997), Speech production learning in adults with chronicdevelopmental stuttering, in W. Hulstijn, H.F.M. Peters & P.H.H.M. van Lieshout (Eds.), (op.cit.), 221-232.

McClean M.D. (1990), Neuromotor aspects od stuttering: Levels of impairment and disability, inJ.A. Cooper, Research needs in stuttering: Roadblocks and future directions, ASHA, Rockville,64-71.

McClean, M.D. (1997), Functional components of the motor system: An approach to understandingthe mechanisms of speech disfluency, in W. Hulstijn, H.F.M. Peters & P.H.H.M. van Lieshout(Eds.), (op. cit.), 99-118.

McClean, M.D., Goldsmith, H. & Cerf, A. (1984), Lower-Lip EMG and displacement duringbilabial disfluencies in adult stutterers, J. Speech Hearing Res., 27, 342-349.

McClean, M.D., Lewandowski, D.R., Cord, M.T. (1994), Intersyllabic movement timing in thefluent speech of stutterers with different disfluency levels, J. Speech Hearing Res., 37, 1060-1066.

MacKay D.G. (1987), The organization of perception and action: A theory for language and othercognitive skills, Springer, New York.

MacNeilage, P.F. (1970), Motor control of serial ordering of speech, Psychological Reviews, 77,

182-196.Magno Caldognetto, E., Zmarich, C., Bettini, F. & Ferrigno, G. (1997), Simoultaneous analysis of

lips, jaw and tongue movements with an integrated optical tracking and EPG system, in W.,Hulstijn, H.F.M., Peters & P.H.H.M., van Lieshout (op. cit.), 439-444.

Neilson M.D. & Neilson P.D. (1987), Speech motor control and stuttering: A computational modelof adaptive sensory-motor processing, Speech Communication, 6, 325-333.

Nespor M. (1993), Fonologia, Il Mulino, Bologna.Perrier, P., Ostry, D.J., Laboissiere, R. (1996), The equilibrium point hypothesis and its application

to speech motor control, J. Speech Hearing Research, 39, 365-378.Peters, H.F.M. & Hulstijn, W. (1987), Speech motor dynamics in stuttering, Springer-Verlag, Wien.Peters H.F.M., Hulstijn w. & Starkweather C.W. (1989), Acoustic and phisiological reaction times

of stutterers and nonstutterers, J. Speech Hearing Res., 32, 668-680.Peters, H.F.M., W., Hulstijn & Starkweather, C.W. (1991), Speech motor control and stuttering,

Excerpta Medica, AmsterdamPort R.F. & T. van Gelder (1995), Mind as motion, The MIT Press, Cambridge, MA.Postma A. & Kolk H. (1993), The covert repair hypothesis: prearticulatory repair processes in

normal and stuttered disfluencies, J. Speech Hearing Res., 36, 472-487.Roelofs A. (1992), A spreading activation theory of lemma retrieval in speaking, Cognition, 42,

107-42.Rubin, P.E., Baer, T., Mermelstein, P. (1981), An articulatory synthesizer for perceptual research, J.

Acoust. Am., 70, 321-328.Saltzman, E. (1999), Nonlinear dynamics of temporal patterning in speech, in “Eppur si muove”.

Dynamics of the production and perception of speech, East Bay Institute for Research andEducation, Inc., Martinez, CA, 59-65.

Saltzman, E.L. & Munhall, K.G. (1989), A dynamical approach to gestural patterning in speechproduction, Echological Psychology, 1, 333-382.

Siegel, G. (1998), Stuttering: Theory, Research, and Therapy, in A.K. Cordes & R.J. Ingham (eds.),Treatment efficacy for stuttering, Singular Publishing Group, San Diego, 1998, 103-113.

Schmidt, R.A. (1988), Motor control and learning: A behavioral emphasis, Human Kinetics,Champagne, IL.

Shriberg, L.D. & Lof, G.D. (1991), Reliability studies in broad and narrow phonetic transcription,Clinical and Linguistic Phonetics, 5, 225-279.

Smith A. Dynamic interactions of factors that impact speech motor stability in children and adults.In W. Hulstijn, H.F.M. Peters & P.H.H.M. van Lieshout (op. cit.), 143-150, 1997.

Smith, A.& Kelly, E. (1997), “Stuttering: a dynamic, multifactorial model”, in R.F., Curlee & G.M.Siegel, Nature and treatment of stuttering. New directions, Allyn & Bacon, Boston, 218-235.

Smith A., Goffman L., Zelaznik H., Ying G. & McGillem (1995), “Spatiotemporal stability andpatterning of speech movement sequences”, Experimental Brain Research, 104, 493-501.

Stager, S.V., Denman, D.W., Ludlow, C.L. (1997), “Modifications in aerodynamic variables bypersons who stutter under fluency-evoking conditions”, J. Speech Hearing Res., 40, 832-847.

Stemberger J.P. (1985), The lexicon in a model of language production, New York, Garland.Sternberg , S., Knoll, R.L., Monsell, S., & Wright, C.E. (1978), Motor programs and hierarchical

organization in the control of rapid speech, Phonetica, 45, 145-197.Story, R.S., Alfonso, P.J., Harris, K.S (1996), “Pre- and posttreatment comparison of the kinematics

of the fluent speech of persons who stutter”, J. Speech Hear. Res., 39, 991-1005.Starkweather, C.W. (1987), “Fluency and Stuttering”, Prentice-Hall, Englewood Cliffs.Starkweather, C.W. (1995), A simple theory of stuttering, J. Fluency Dis., 20, 91-116.Van Lieshout, P.H.H.M. (1995), Motor planning and articulation in fluent speech of stutterers and

nonstutterers, Nijmegen Institute for Cognition and Information, University Press Nijmegen,The Netherlands.

Weismer G., Tjaden, K. & Kent, R.D. (1995), Can articulatory behavior in motor speech disorders

be accounted for by theories of normal speech production?, J. Phonetics, 23, 149-164.World Health Organization (1977), Manual of the international statistical classification of

deseases, injuries and cause of death, Vol. 1, Geneva.Yaruss, J.S. (1998), Describing the consequences of disorders: Stuttering and the International

Classification of Impairments, Disabilities, and Handicaps, J. Speech Hear. Res., 41, 249-257.Zelaznik, H.N., Smith, A., Franz, E.A. & Ho, M. (1998), Differences in bimanual coordination

associated with stuttering, Acta Physiologica, 96, 229-243.Zmarich, C. (1991), Una revisione critica degli studi 'linguistici' sulla balbuzie, Acta Phoniatrica

Latina, XIII, 495-514.Zmarich, C. (1998), La balbuzie come disturbo della produzione articolatoria, in Giornata sulla

balbuzie: contenuti e metodi nella valutazione e nel trattamento logopedico, Venezia, Pistellato,3-60.

Zmarich, C., Magno Caldognetto, E., Vagges, K. (1994), La balbuzie come disturbo dellaproduzione articolatoria, Acta Phoniatrica Latina, 16, 157-183.

Zmarich, C., Magno Caldognetto, E., Vagges, K. (1995), Articulatory kinematics of lips and jaw inrepeated /pa/ and /ba/ sequences in Italian stutterers, Proceedings of the First World Congress onFluency Disorders, Munich, vol. 1, 43-47.

Zmarich, C. & Magno Caldognetto, E. (1997), Analysis of lips and jaw multi-peaked velocity curveprofiles in the fluent speech of stutterers and nonstutterers, in W., Hulstijn, H.F.M., Peters &P.H.H.M., van Lieshout (op. cit.), 177-182.