Apraxia of speech is

a phonetic-motoric disorder of speech production caused by inefficiencies in the translation of a well-formed and filled phonologic frame to previously learned kinematic parameters assembled for carrying out the intended movement, resulting in intra-and inter-articulator temporal and spatial segmental and prosodic distortions. It is characterized by distortions of segments, intersegment transitionalization resulting in extended durations of consonants, vowels and time between sounds, syllables and words. These distortions are often perceived as sound substitutions and the misassignment of stress and other phrasal and sentence-level prosodic abnormalities. Errors are relatively consistent in location within the utterance and invariable in type. It is not attributable to deficits of muscle tone or reflexes, nor to deficits in the processing of auditory, tactile, kinesthetic, proprioceptive, or language information. (McNeil, Robin, and Schmidt, 1997, p. 329)

The kernel perceptual behaviors that differentiate apraxia of speech (AOS) from other motor speech disorders and from phonological paraphasia are (1) lengthened segment (slow movements) and intersegment (segment segregation) durations (overall slowed speech), resulting in (2) abnormal prosody across multisyllable words and phrases, with a tendency to make errors on more stressed than unstressed syllables; (3) relatively consistent trial-to-trial location of errors and relatively nonvariable error types; and (4) impaired measures of coarticulation. Although apraxic speakers may produce a preponderance of sound substitutions, these substitutions do not serve as evidence of either AOS or phonemic paraphasia. Sound distortions serve as evidence of a motor-level mechanism or influence in the absence of an anatomical explanation; however, they are not localizable to one part of the motor control architecture and, taken alone, do not differentiate AOS from the dysarthrias. Acoustically well-produced (nondistorted) sound-level serial order (e.g., perseverative, anticipatory, and exchange) errors that cross word boundaries are not compatible with motor planning-or programming-generated mechanisms and are attributable to the phonological encoding mechanism.

Although this motor speech disorder has a languorous and tortuous theoretical and clinical history and is frequently confused with other motor speech disorders and with phonemic paraphasia, a first-pass estimate of some of its epidemiological characteristics has been presented by McNeil, Doyle, and Wambaugh (2000).

Based on retrospective analysis of the records of 3417 individuals evaluated at the Mayo Clinic for acquired neurogenic communication disorders, including dysarthria, AOS, aphasia, and other neurogenic speech, language, and cognitive disorders, Duffy (1995) reported a 4.6% prevalence of AOS. Based on this same retrospective analysis of 107 patient records indicating a diagnosis of AOS, Duffy reported that 58% had a vascular etiology and 6% presented with a neoplasm. One percent presented with a seizure disorder and 16% had a diagnosis of degenerative disease, including Creutzfeldt-Jakob disease and leukoencephalopathy (of the remaining, 9% were unspecified, 4% were associated with dementia, and 3% were associated with primary progressive aphasia). In 15% of cases the AOS was traumatically induced (12% neurosurgically and 3% concomitant with closed head injury), and in the remaining cases the cause was undetermined or was of mixed etiology. Without doubt, these proportions are influenced by the type of patients typically seen at the Mayo Clinic, and may not be representative of other patient care sites.

Among all of the acquired speech and language pathologies of neurological origin, AOS may be the most infrequent. Its occurrence unaccompanied by dysarthria, aphasia, limb apraxia, or oral-nonspeech apraxia is extremely rare. Comorbidity estimates averaged across studies and summarized by McNeil, Doyle, and Wambaugh (2000) indicated an AOS/oral-nonspeech apraxia comorbidity of 68%, an AOS/limb apraxia comorbidity of 67%, an AOS/limb apraxia and oral-nonspeech apraxia comorbidity of 83%, an AOS/aphasia comorbidity of 81%, and an AOS/dysarthria comorbidity of 31%. Its frequent co-occurrence with other disorders and its frequent diagnostic confusion with those disorders that share surface features with it suggest that the occurrence of AOS in isolation (pure AOS) is extremely rare.

The lesion responsible for AOS has been studied since Darley (1968) and Darley, Aronson, and Brown (1975) proposed it as a neurogenic speech pathology that is theoretically and clinically different from aphasia and the dysarthrias. Because Darley defined AOS as a disorder of motor programming, the responsible lesion has been sought in the motor circuitry, especially in Broca's area. Luria (1966) proposed that the frontal lobe mechanisms for storing and accessing motor plans or programs for limb gestures or for speech segments were represented in Broca's area. He also proposed the facial region of the postcentral gyrus in the parietal lobe as a critical area governing coordinated movement between gestures (speech or nonspeech). AOS-producing lesions subtending Broca's area (Mohr et al., 1978) as well as those in the postcentral gyrus (Square, Darley, and Sommers, 1982; Marquardt and Sussman, 1984; McNeil et al., 1990) have received support. Retrospective studies of admittedly poorly defined and poorly described persons purported to have AOS (e.g., Kertesz, 1984) do not show a single site or common cluster of lesion sites responsible for the disorder. Prospective studies of the AOS-producing lesion have been undertaken by a number of investigators. Deutsch (1984) was perhaps the first to conduct a prospective search, with a result that set the stage for most of the rest of the results to follow. He found that 50% of his AOS subjects (N = 18) had a lesion in the frontal lobe and 50% had posterior lesions. Marquardt and Sussman's (1984) prospective study of 12 subjects with AOS also failed to reveal a consistent relationship among lesion location (cortical versus subcortical, anterior versus posterior), lesion volume, and the presence or absence of AOS. Dronkers (1997) reported that 100% of 25 individuals with AOS had a discrete left hemispheric cortical lesion in the precentral gyrus of the insula. One hundred percent of a control group of 19 individuals with left hemispheric lesions in the same arterial distribution as the AOS subjects but without the presence of AOS were reported to have had a complete sparing of this specific region of the insula. McNeil et al. (1990) reported computed tomographic lesion data from four individuals with AOS unaccompanied by other neurogenic speech or language pathologies. The only common lesion site for these “pure” apraxic speakers was in the facial region of the left postcentral gyrus. Two of the four AOS subjects had involvement of the insula, while two of three subjects with phonemic paraphasia (diagnosed with conduction aphasia) had a lesion in the insula. Two of the four subjects with AOS and one of the three subjects with conduction aphasia evinced involvement of Broca's area. The unambiguous results of the Dronkers study notwithstanding, the lesions responsible for AOS remain open to study. It is clear, however, that the major anterior/posterior divisions common to aphasiology and traditional neurology as sites responsible for nonfluent/fluent (respectively) disorders of speech production are challenged by the AOS/lesion data that are available to date.