Clinical manifestations

Presentation and course

John Hughlings Jackson's preference for subdividing aphasia into fluent and nonfluent subtypes (162) remains valid (186), because ease of speech production is the most distinguishing feature among individuals with diverse aphasic presentations (178; 181; 01). Regardless of fluency ability, grammar, repetition, naming, and comprehension are usually impaired among the perisylvian aphasias.

Fluency disturbance may appear in either of two forms: (1) reduced output, or (2) excessive output. Nonfluent speech is characterized by a struggle to speak or frustration with deficient production. In addition, nonfluent speech may be dysprosodic. The nonfluent individual frequently shows oral apraxia or weakness. Some patients increase gesturing with their upper extremities to overcome their fluency impairments (06).

In severe nonfluent aphasia, patients may still produce words. But "to utter words is not necessarily to speak" (162; 163); thus, the expression of an original thought or proposition may be impaired despite intact speech output, as with a parrot. Certain conditions may allow nonfluent aphasics greater ease and range of expression. Visually stimulated discourse, such as when patients are asked to describe picture sequences, can result in greater fluency than during minimally cued conversation (104). Emotional upset may allow them to swear capably (162; 163). Some nonfluent patients can sing words to familiar songs (352). Nonetheless, such retained speech is generally "nonpropositional," that is, reflecting the liberation of well-learned expressions, rather than the elaboration of new phrases. However, in most aphasic patients, nonpropositional speech is as likely to be impaired as is propositional speech, with a few exceptions (208). Excessive speech is less often described and has been termed hyperlalia. This may be seen following right hemisphere perisylvian lesions (353; 63).

Grammatical disturbances may be expressive or receptive. Nonfluent aphasia may involve conspicuous agrammatic speech, in which primarily nouns and verbs are expressed with relatively few adjectives, adverbs, and prepositions. The effect is similar to the speech characteristics of a foreigner who is not conversant in the local language, or "Tarzan." In addition, root words may not be properly modified for their roles within sentences to reflect such distinctions as singular versus plural, or present versus past or future occurrences.

Impaired grammatical comprehension may appear when patients encounter sentences that do not comprise the familiar subject-verb-object sequence or that have elaborate grammatical constructions; thus, agrammatic patients may fail to comprehend who died in the statement “the lion was killed by the leopard," or to deduce the sex of "my brother-in-law's uncle's daughter." Agrammatism, therefore, is an aphasic disorder that disrupts the application of language-specific rules, rather than impeding the meaning, reconstitution, or sound processing of individual words.

Repetition and naming are similar processes in that the patient must generate a specific word or phrase that corresponds to an external stimulus or a pre-verbal thought. In repetition, the patient must acoustically analyze another individual's speech to produce the identical expression. Recognizing the spoken utterance benefits repetition. In naming, the individual must retrieve a previously learned word when either sensing an object (eg, touching it) or conversing (eg, retrieving the correct word for a specific definition).

Two kinds of deficit may appear on these activities:

Absent response. On naming tasks, this is termed anomia and may involve considerable effort and "word finding difficulty." This is essentially a severe form of the commonly experienced "tip-of-the-tongue" phenomenon (54), in which the person is aware of the desired concept but falls short producing the exact word. Patients whose language problem is primarily the failure to repeat or name are usually well aware of their deficits. Anomia, of course, may appear in other cognitive disorders, including impairments of memory, attention, or self-initiation.

Incorrect response (paraphasia). There are three kinds of paraphasias: (1) Phonemic, or literal, paraphasias, in which the response differs from the correct word by one letter or sound, such as saying "shammer" for "hammer." (2) Semantic paraphasias, in which the wrong word is produced, one that is usually related to the target. For example, "pliers" for "hammer." (3) Neologisms, in which the response does not occur in the speaker's language. For example, "humdup" for "hammer." Neologisms, although strange, generally reflect the normal sounds and rules for combining sounds within the patient's language and, therefore, do not seem to come from another language. This condition must be distinguished from the “foreign accent” syndrome following brain injury, in which patients speak consistently as if sounding as if from a foreign country (330). This disturbance, a form of aprosodia, alters the production of speech sounds consistently, in contrast to aphasic patients.

The repetition of test stimuli during the examination of aphasic patients may considerably depend on the context. In a fascinating study, patients who were diagnosed with Wernicke aphasia were more impaired during formal repetition testing than when they were asked to search for specific objects and to say their names while doing so (252). This suggests that language in aphasic individuals is particularly sensitive to task familiarity.

More familiar tasks (such as saying the name of an object while looking for it) are associated with improved speech than are procedures that are associated with standard clinical evaluations.

Ironically, aphasia often includes disinhibited self-repetition, despite the frequent failure to repeat during formal testing. Perseveration is not uncommon. In particular, aphasic patients may occasionally give the same reply to different questions (164), which is termed "recurrent perseveration" (284). A more severe kind of disinhibited self-repetition that may appear in aphasia is known as recurring utterances (264), stereotypies (243), or speech automatisms (42). This may have been the primary disturbance of Paul Broca’s landmark patient in the 1860s who was nicknamed “Tan” because this single word was a recurring utterance (293). In this disturbance, the patient's speech almost exclusively consists of a single word, phrase, nonsense word, or even simply a consonant-vowel pair, such as "mi." The element is often repeated many times consecutively as if in a phrase ("mi mi mi mi"). Regardless of its nature, the specific perseverated expression in recurring utterances generally appears to have no particular significance to the patient. Why aphasic patients may become "stuck" on the particular expression is unknown. Most patients with recurring utterances have global aphasia, but this phenomenon also appears in less severely aphasic patients, primarily in Broca aphasia (243; 42; 276).

Some unusual variations of recurring utterances have been reported. In "moriatic aphasia," the patient excessively produces diverse nonverbal vocalizations (127). In "semistereotypic speech," the repeated item may be combined with different words each time the patient speaks. This has been reported only among Japanese speakers, which may result from the almost exclusively Japanese practice of affixing content words (nouns or verbs) to functors (eg, particles and auxiliary verbs) (139).

Aphasia may disrupt manual as well as verbal forms of communication. It is not unusual for individuals with aphasia to have impaired writing as well. However, in many such instances, impaired fine motor control of the dominant hand can interfere with assessing aphasic disturbances in writing. This limitation can be bypassed in rare instances of crossed aphasia, in which a stroke resulting in aphasia occurs in the hemisphere opposite to the one that is primarily involved with hand control or writing. Such rare cases can exhibit distinct aphasic disturbances in writing (95).

Commonly, the disruption of speech and written communication in aphasia occur to about the same extent. Nonetheless, exceptions occur, such as reading comprehension better than speech comprehension, or speech expression better than written expression (26; 297). Such dissociations suggest that mechanisms for symbolic communication via speaking and writing are independent from each other at some level.

The increasing use of keyboard entry on computers or other electronic devices to communicate rapidly, which minimizes the difficulty with hand control, will likely increasingly reveal, or at least raise suspicion of, aphasic disturbances. For example, “dystextia,” or the impaired composition or completion of instant messages via keyboard entry, has been reported following stroke or migraine (68; 348; 273; 58). Because text entry involves more than linguistic skill (eg, manual coordination, spatial attention, and visual and tactile perception), abnormal text entry does not by itself imply aphasia. However, text entries can reveal signs associated with aphasia, such as agrammatism, pronounceable neologisms, or jargonagraphia (273).

It is, therefore, not surprising that aphasia can also disrupt sign language, either its production or comprehension (82).

Comprehension failure may result either from impaired grammatical (or syntactic) comprehension (ie, in relation to other words in the sentence or phrase), or from impaired semantic comprehension. The latter affects comprehension of individual word meaning. Patients with impaired semantic comprehension may understand what kind of response is expected (eg, answer yes or no, or make a specific movement), but they respond incorrectly (eg, if asked to raise their hands, they may stick out their tongues). They may nod when asked whether a rock can float. Impaired comprehension may appear gradually during examination, rather than occurring at the outset (121), suggesting a kind of cognitive fatigue, although the patient remains attentive. Speech comprehension may fluctuate even though the examiner repeats the same words (38; 166). Accordingly, the examiner's pausing during speech may improve comprehension (205).

Studies with electrophysiologic evaluations suggest that subconscious speech discrimination may still occur despite overt comprehension failure (07; 100). Indeed, auditory evoked response testing can indicate that a differential P3 wave can be recorded when patients with minimal consciousness or persistent vegetative state following severe brain injury hear their own name amidst a series of other familiar names (260). This finding indicates that semantic processing of items that are highly relevant to patients can resist severe brain injury. In turn, this suggests that the absence of overt behavioral changes to speech in aphasic patients cannot be used to infer that the patient is incapable of preferentially recognizing meaningful words. Thus, care should be taken when talking to other individuals about the patient in front of the patient.

Aphasia subtypes. Some clinicians feel that fluent and nonfluent aphasias differ in the quality of the foregoing deficits. However, aphasia research, at least following stroke, fails to control many important variables (350), including lesion extent, time following illness onset, hand preference, educational and cultural background, prior cerebral injury, age, practice or familiarity effects, current medications, and extent of speech rehabilitation. Additionally, the definitions of aphasia subtypes are not universally agreed on and are inconsistent (291; 61). Most aphasic patients do not fulfill the criteria for such specific aphasic subtypes beyond the nonfluent versus fluent classification (04; 324; 132; 250). Interindividual differences may also be considerable within aphasia types. Individual capacities for recovering from aphasia may differ greatly (06). Thus, one cannot confidently describe the qualitative differences among aphasia subtypes.

One of the most popular classification schemes is influenced by neuropathological correlation and subdivides perisylvian aphasias into the following disorders: Broca aphasia, conduction aphasia, Wernicke aphasia, and global aphasia. Although these terms are not applied consistently or defined precisely, they are useful for roughly indicating the language impairments among patients and they, therefore, allow rapid communication among clinicians.

One large stroke study found that the leading aphasia subtypes after stroke were Broca, anomic, and global aphasia, accounting for about three-quarters of all aphasias (157).

Broca aphasia is a nonfluent aphasia with agrammatism, repetition failure, and relatively retained comprehension, usually accompanied by right hemiparesis and oral apraxia.

Nonetheless, comprehension is impaired, although usually more for grammatical relationships than for individual word meanings (282; 316). It is thought to emerge only in recovery from severe aphasia (235), although some authors believe that it may appear acutely following brain injury (04). Rarely, repetition may be preserved in this disorder (321).

Conduction aphasia (termed because it was thought to result from impaired signal conduction between Wernicke's and Broca's areas in the left hemisphere) is a fluent aphasia typified by phonemic paraphasia and impaired repetition, but with relatively preserved speech comprehension. Conduction aphasics are well aware of their deficits, and try repeatedly to correct their speech errors through minor modifications of single words, a process termed conduite d'approche (57); however, dedicated self-correction occurs in other aphasic subtypes as well (96). Conduction aphasia may evolve from Wernicke aphasia.

Wernicke aphasia is also a fluent aphasia, but with severely impaired comprehension. In contrast to the agrammatism of Broca aphasia, the speech in Wernicke aphasia utilizes all parts of speech, but tends to express ideas in an elaborate, roundabout manner with nonspecific nouns and verbs. An example typical of such empty expression would be: "I went to the thing that gives you the trip around and around again and came back and did the whole thing again." Speech may be pressured (hyperlalia) and include neologisms. The patient often is unaware of the disturbances. However, some patients with fluent, nonsensical speech that is typical of Wernicke aphasia may nonetheless have minimally impaired comprehension (182). Patients with Wernicke aphasia may also have alexia; they typically do not have hemiparesis.

Wernicke aphasics may acutely fear or become upset with other individuals due to their impaired comprehension and insight (277). Consequently, they may resemble psychotic patients in their agitation and jargon. Such agitation may be unintentionally provoked by clinicians who fail to recognize the disorder.

A less described aphasia subtype involves the select impairment of the comprehension of metaphors, or figurative language, despite normal scoring on standard language assessment (the Western Aphasia Battery). In the single test that has been developed, participants are given a multiple choice reading test to select the one phrase that best matches the meaning of a sentence that relies on a figurative, rather than literal, understanding (eg, “The debate spun into a brawl,” when there was literally no spinning). Patients with either left or right hemisphere vascular injury were vulnerable to this difficulty compared to healthy controls (161). It is not yet clear whether this mild a language disturbance affects real-world behavior.

Global aphasia (otherwise known as severe aphasia or total aphasia) usually appears as a nonfluent disorder. Essentially, all language functions are severely affected, so that the patient appears mute or has markedly reduced output. Severe right hemiparesis is usually the rule, although exceptions occur (19). Although speech comprehension is severely affected, such patients usually respond appropriately to emotional intonation or facial expression; thus, communication on a basic level is retained. Nonetheless, patients often nod indiscriminately when they are asked questions, which may lead clinical staff or family members to misjudge the level of comprehension. Based on the memoirs of one individual who recovered from global aphasia, thought is still present, but is not coded in words or "inner speech" (239). This may explain why this author's memory was not as detailed as it was premorbidly.

As indicated above, some global aphasic patients have recurring utterances and, therefore, have a fluent, rather than a nonfluent disorder. Recurring utterances may also rarely appear in patients who otherwise qualify as having either Broca or Wernicke aphasia (42; 276).

Prognosis and complications

Aphasia is generally disabling. However, it may actually acutely be associated with improvement in some functions, such as lie detection (94) and jigsaw puzzle solving (137). Prognosis depends on the pathology. Generally, poststroke aphasia generally improves in the absence of recurrent stroke or other subsequent brain disease (194), even though the boundary of cerebral infarction itself may expand over years (244).

Stroke-induced aphasia is associated with twice the mortality from stroke as compared with nonaphasic stroke (36% vs. 16%). This likely reflects underlying cardiovascular pathology or general stroke severity, rather than due to aphasia itself (194; 90). In aphasia recovery, clinical subtype may change, such that Wernicke aphasia may resolve to conduction aphasia, for example (97). Indeed, most aphasia subtypes lessen in frequency, except that the incidence of conduction aphasia increases following stroke (194), probably reflecting the evolution of subtypes. Initial stroke severity is related to the extent of recovery, such that more severe aphasia has the worst prognosis (258). By 18 months, about half of initially affected patients are appreciably aphasic. Factors associated with aphasia recovery in the chronic state include small lesion size, young onset, and higher educational attainment (315). Nonetheless, considerable inter-individual differences affect outcome, such that it is impossible to predict an individual's extent of recovery simply by diagnosing the aphasia subtype (202; 27). In persons with mild to moderate aphasia, the initial aphasia severity highly correlates with outcome at three months (197). Larger infarcts are generally associated with less recovery (133; 257). Socioeconomic status does not influence outcome (323).

Although stroke is generally a non-progressive illness that is followed by improvement, a new entity is stroke following the COVID-19 illness that induces infarction. Because of the association between cerebral infarction with the COVID-19 illness, it is not surprising that aphasia could follow from the illness. In the earliest published such cases, conduction aphasia and Broca aphasia have been reported, respectively (267; 302). At present, the natural history following from cerebral infarction from the COVID-19 illness has not been determined.

Broca aphasia tends to improve substantially more in general language functions than do other kinds of perisylvian aphasia (17). Lesion volume correlates with extent of recovery (243). In some cases, nonfluent aphasia recovers to a purely articulatory disorder without language disturbance (289). Nonetheless, the variability within individuals with respect to lesion size and aphasia type does not allow the reliable prediction of recovery at the individual level (195).

Impaired speech comprehension appears to affect the level of functional independence after stroke more than do other language disturbances (128). Acute speech comprehension ability strongly predicts discharge destination after hospitalization, such that impaired patients have a greater likelihood of chronic institutionalization (134). Nonetheless, occasionally patients with chronically impaired speech comprehension can be functionally or even vocationally independent on nonspeech tasks (24; 218; 66; 176; 173; 27; 223; 62). Speech comprehension disturbance for single words depends on stroke lesion volume (296). However, there is a minimal relationship between lesion volume and recovery from impaired comprehension of sentences, except at extremes of lesion volumes (295). Recovery from impaired comprehension of single words is generally good by six months. In contrast, involvement of the superior temporal gyrus is associated with poor prognosis for recovery from impaired sentence comprehension, whereas lesions that spare the temporal lobe are associated with improved prognosis.

Anomia normalizes by six months in only one third of patients following left cerebral infarction (185). Lesion volume correlates with extent of recovery.

Global aphasia improves maximally during the initial six months, but usually does not show substantial improvement, even years later (247) or following intensive therapy (286). Nonetheless, a possibly exceptional case of extended language recovery in global aphasia has been reported, demonstrating continual improvement over a span of 25 years, leading to improved spontaneous speech (307). Language assessment within the first few weeks may more reliably indicate prognosis one year later than do initial neuroimaging (216). Speech production is more likely to improve than comprehension (285). Lesion extent correlates with recovery (245). Prognosis may be better among patients with strictly subcortical lesions (102). Recovery among patients with global aphasia without hemiparesis can vary, but is generally poor (144). Global aphasia is generally associated with reduced recovery of basic activities of daily living in comparison to other aphasia subtypes (253).

Clinical vignette

A 69-year-old, right-handed man was found by friends on the floor of his home, appearing confused. The duration of confusion was unknown, but he had been seen interacting normally 48 hours earlier. Past medical history documented several years of atrial fibrillation and alcohol abuse. He was taking no medications at the time. He had a 10th grade education. Examination at the hospital disclosed a talkative, obese man who did not appear to comprehend the examiner well. He had a rapid, irregular pulse and mildly elevated blood pressure, along with superficial abrasions on his right extremities and a grade II/VI cardiac systolic ejection murmur.

His speech was fluent with frequent paraphasic errors and neologisms. He could only comprehend body midline commands (eg, "Close your eyes"). His writing was illegible. The remainder of neurologic testing was notable for decreased pin responsiveness on his right side and mild proximal right upper extremity weakness. The visual fields appeared to be full.

Cranial CT scan disclosed a large left posterior temporal lobe lucency that was interpreted to be an acute stroke.

Examination 11 days after admission disclosed considerably improved speech comprehension, but nonetheless he was mistaken when indicating some body parts and replied to some questions erroneously. His speech still consisted of neologisms and semantic and phonemic paraphasias; he seldom attempted to correct himself.

The clinical impression by this time was fluent aphasia, consistent with Wernicke aphasia.

Re-evaluation three months later indicated significant reduction in neologisms. In contrast with the earlier evaluation, he frequently corrected his speech, which still contained semantic and phonemic paraphasic errors and word-finding difficulty.

The clinical impression was that Wernicke aphasia had resolved to conduction aphasia.

Biological basis

Anatomic localization

Aphasia is most commonly noted following left hemisphere injury. Perisylvian tumors cause less severe aphasic disturbances than do strokes in the same region (08). Individuals who show traditionally-defined aphasia following right hemisphere injury are regarded as having atypical cerebral hemispheric functional organization. Nonetheless, some mild language disorders often accompany right hemisphere injury in individuals thought to have typical cerebral lateralization, including disturbances affecting comprehension at the paragraph level and appreciation of abstract meaning. It is debatable whether these disturbances are aphasic or attentional. However, detailed grammar testing can identify acquired language impairments following either left or right hemispheric infarction, showing that some kinds of language processing are not specific to the left hemisphere (217).

Acute nonfluent aphasia occurs most often with anterior lesions (222; 103; 190). CT in chronic nonfluent aphasia, following left hemisphere stroke, usually demonstrates lesions that are centered about the Rolandic fissure (central sulcus) (186). Contrary to classical teaching, Broca’s area in the opercular frontal cortex is not an obligatory lesion site (152). Indeed, lesion restricted to Broca’s area may actually result in impaired speech initiation (transcortical motor aphasia) without agrammatism or repetition failure (06). Hence, the clinical justification of the term "Broca's area" is unfounded.

Although not typically considered to reflect aphasia, the decrease with naming ability that normally occurs with aging—the “tip-of-the-tongue” phenomenon that includes frustration with inability to produce the intended word—is associated with progressive atrophy of the left insular cortex (301).

Hyperlalia or excessive fluency may appear following right hemisphere injury (187; 353; 63), with the patient often appearing apathetic and indifferent. Alternatively, hyperlalia may result from posterior left hemisphere injury, in which case its coexistence with impaired comprehension suggests Wernicke aphasia.

Chronic speech comprehension disturbance for phrases typically involves the left posterior superior temporal lobe, or Wernicke’s area (219). Nonetheless, in many instances, Wernicke’s area is spared when comprehension of single words is impaired, and conversely, lesion of this area may not impair comprehension of single words (296). However, one study found in a small sample of acute stroke patients that hypoperfusion of Brodmann area 22 (equated with Wernicke’s area by the authors) was exclusively associated with impaired single word comprehension, which could be reversed when phenylephrine was used to reperfuse the area (155). There is no consensus on the precise boundaries of Wernicke’s area (43) and, therefore, the term is more historically interesting than functionally useful, similar to Broca’s area. Comprehension for phrases and sentences, as opposed to single words, may be disrupted by lesions in the frontal lobe (06).

The most severe and enduring repetition disturbances following stroke are associated with lesions of "Wernicke's area" (294). Phonemic paraphasias typically follow acute injury to subcortical pathways that link frontal and temporoparietal areas (06; 281). Semantic paraphasias, anomia, and agrammatism are not precisely localized following left perisylvian injury (233; 06). Limited evidence suggests that anterior left perisylvian injury is associated with persistent impaired syntactic comprehension (324). Transient anomia may be induced by electrical cortical stimulation, most often in the left superior temporal gyrus. This is less common when tumor invades the temporal lobe. Although it is not clear how electrically-induced anomia anatomically compares with stroke-induced anomia, surgically-induced aphasia is less likely to occur the more distant the surgery is from such electrically-identified "language sites" (141).

Global aphasia is most commonly seen with extensive perisylvian injury. Typically, such individuals have marked right hemiparesis as well; however, global aphasia may also appear with minimal or no hemiparesis, either following two separate lesions affecting anterior and posterior perisylvian cortex (332; 200; 325) or an acute solitary lesion (101; 44; 144).

Recurring utterances have been specifically associated with acute lesion of the white matter laterally adjacent to the frontal horn of the left lateral ventricle (the subcallosal fasciculus) or the mid-portion of the left periventricular white matter (243). There appears to be no significant difference between lesion sites associated with recurring utterances and with global aphasia (264; 243); therefore, lesion localization in these cases does not account for why some patients are nonfluent, whereas others have recurring utterances.

Pathophysiology

Interhemispheric differences and interactions. Experimental findings in healthy individuals (eg, positron emission tomography scans and dichotic listening studies) overwhelmingly implicate the left hemisphere in language processing (263). The neonatal brain typically begins development without a left hemisphere bias in language processing (259), and the left Sylvian fissure subsequently elongates as the infant acquires proficiency with language (130). Although research suggests some language functions may normally occur in the right hemisphere, particularly among some non-right-handed individuals (23; 317), and as suggested by the rare instances of crossed aphasia (14; 77), the prevalence of aphasia (or at least the most severe language disturbances) following left hemisphere injury suggests that the left hemisphere is preferentially involved in language processing, at least for basic aspects of propositional language. Therefore, researchers have attempted to characterize the hemispheric differences in anatomy and signal processing among right-handers to understand fundamental language mechanisms and, hence, the vulnerability of left brain injury toward producing aphasia.

Interhemispheric anatomical asymmetries may reflect left hemisphere specialization for language. The Sylvian fissure is longer in the left hemisphere than in the right in most individuals. This asymmetry is less pronounced in chimpanzees and is absent among rhesus macaques (354; 59). In addition, humans have a greater proportion of white matter to grey matter in the left hemisphere than, to a progressively lesser extent, apes and monkeys (306). These differences may suggest part of the basis for the apparently greater ability of chimpanzees than monkeys to master human-engineered communication, such as selecting and sequencing handheld symbolic shapes, pressing symbol-encoded response keys, or using sign language (255). In addition, a crucial difference between humans and other primates is superior auditory long-term memory, which may account for the greater faculty for language among humans (117).

Quantitative assessment of fiber tract morphology indicates that the left arcuate fasciculus is larger and has a greater degree of anisotropic molecular movement along its bundle than the right hemisphere in humans (266). This suggests that the left arcuate fasciculus is more active than the right and is consistent with left hemispheric specialization for language. Functional neuroimaging research suggests that specific brain areas in the left hemisphere are dedicated to linguistic processing, as opposed to sharing cortical space with other cognitive functions (98). Aphasia recovery can be associated with improved structure of the left hemisphere arcuate fasciculus as found by diffusion tensor imaging on MRI scanning, though this is not a consistent finding in stroke patients (13).

The intracarotid amobarbital (or Wada) test among epileptic patients induces aphasia most commonly after left cerebral anesthesia. Several studies have indicated in such subjects that certain left hemisphere cortical structures are larger than the right hemisphere homologs, including part of Broca’s area, the planum temporale, and the occipital lobe's length (70; 111; 112). Such asymmetries are less marked or reversed among individuals with right hemisphere language specialization. These findings suggest that anatomical specialization for basic language is associated with relative enlargement of contributing anatomical structures.

The left hemisphere may preferentially analyze information sequentially (204), whereas the right hemisphere is biased toward rapid complex pattern processing (40). Other studies suggest that the left hemisphere has primarily spatially narrow visual processing (150) and brief rather than sustained attention span (347; 248). These proclivities toward brief, detail-oriented cognitive processing may favor the left hemisphere for elementary aspects of language, particularly sound processing, syntax, and individual word meaning. Consistent with this idea are the findings that the left temporal lobe is preferentially activated in deaf signers who view sign language, and that deaf signers are more profoundly aphasic after left than right hemisphere injury (328; 153). This suggests that the left temporal lobe is specialized for symbol interpretation, regardless of whether the message is acoustic or visual (210). In addition, left frontal and temporal metabolic activity during repetition is associated with significant corpus callosum hypometabolism, suggesting that interhemispheric neuronal signaling is inhibited during some kinds of speech (170).

In contrast, the right hemisphere may preferentially process language on broader levels than fine sequences. Several lines of research suggest that the right hemisphere in most healthy individuals has limited linguistic ability, but it nonetheless takes part considerably (if not exclusively) in well-learned, "automatic," nonpropositional speech that is produced with minimal or no premeditation, such as greeting, recitation by rote, and swearing (300; 308; 193; 136; 279; 311; 270; 334); however, rote recitation of a lengthy passage such as a pledge is more likely to involve the left hemisphere (46; 41). Consequently, simple automatic expressions are comparatively preserved following left hemisphere injury that results in aphasia. Discourse following right hemisphere injury may not be aphasic in the sense of agrammatism, anomia, phonemic paraphasia, and literal comprehension disturbance, but rather it may show impaired humor and abstraction, tangentiality, and a disposition to metaphor substitution in speech (343; 275; 345; 333). Functional imaging findings implicate the right hemisphere in the processing of multiple meanings during sentence completion tasks (183); thus, it may be said to have a "wide semantic field," in contrast to the left hemisphere. Right hemisphere lesioned patients may also be particularly unable to comprehend the emotional meanings of words (47). Whether these behaviors may be considered linguistic or even aphasic is controversial. On the one hand, such disturbances may reflect a generalized attentional deficit, which is more severe following right hemisphere lesion (78), leaving the individual highly distracted and, therefore, unable to process language abstractly. On the other hand, attentional deficits also occur following left hemisphere injury that produces aphasia (326; 262; 303; 188). It is operationally unclear how one may conclude that the communication deficits following right, but not left hemisphere injury are "attentional."

Aphasia that follows a single cerebral ischemic injury generally improves (97); therefore, the distinguishing features of a patient's aphasia reflect not only disrupted function following infarction of a specific region, but also concurrent compensation from surviving brain regions. A current theory of aphasia recovery posits two different but not mutually exclusive mechanisms: “degeneracy,” the capability of nonlinguistic neural networks to adapt and mediate linguistic processes, and “neuro-displacement,” the disinhibition of latent linguistic neural pathways that are called into action depending on the language demands of the particular linguistic needs (314). The initial impact of left hemisphere stroke that causes aphasia is bilateral hemispheric hypometabolism, followed by more resolution in the right than left hemisphere (64). Several studies have indicated that substantial recovery from aphasia following left hemisphere injury is associated with increased (or at least retained) metabolic activity in brain regions surrounding the injured area (29; 171; 338), whereas poor outcome is associated with increased right hemisphere metabolism (148). Convergent structural evidence comes from study on the integrity of right hemispheric language-analog white matter pathways in left hemisphere stroke patients: the better structural composition of the right hemisphere arcuate fasciculus, as shown by diffusion tensor imaging, the poorer aphasia (or at least impaired word retrieval) recovery following left hemisphere infarction (179). Bihemispheric activation may be associated with greater recovery than strictly right hemisphere activation (60). Activation of the left superior temporal gyrus, at least during word repetition, appears to predict aphasia recovery (149). Nonetheless, this pattern is not consistent across acquired linguistic deficits, because evidence exists supporting right hemisphere compensation following recovery from severe aphasia or alexia from left hemisphere lesion (79; 22; 322; 11; 329), in phonologic processing (238), and in preserving repetition following left hemisphere aphasic injury (35; 318). The specific kind of testing may influence whether more right or left hemisphere activation is found to be involved during recovery. In one study that uses dichotic listening testing and measurement of evoked potentials on a wide variety of linguistic tasks, greater right than left hemisphere activation was seen among recovering aphasic patients (254). Furthermore, indirect evidence suggests that aphasia recovery may shift from the left to the right hemisphere over several months (234). A longitudinal study of cerebral activation with fMRI demonstrated three distinct anatomical patterns during substantial recovery from post-stroke aphasia: (1) reduced left inferior frontal activity within a few days of stroke onset, (2) by two weeks, extensive activation in both hemispheres, especially the right, and (3) by one year, restoration of normal activation patterns, favoring the left hemisphere, associated with nearly normal language (287).

Intrahemispheric differences. Intrahemispheric differences in modeling aphasia pathophysiology have been more controversial than modeling aphasia from interhemispheric differences. That such differences occur is suggested by the relationship between fluency impairment and lesion location within the left hemisphere (181; 103; 190), although exceptions have been noted (144). In addition, phonology and semantics may be independently impaired in left hemisphere disease (156; 292), with phonologic impairments more common following anterior injury.

Broca's and Wernicke's 19th century clinical observations inspired a serial processing model of left hemisphere language function for most right-handers that was highly influential in inspiring current concepts of structural-functional relations in general neuroscience, but is beginning to be regarded largely incomplete (86). The model, extended by Lichtheim (203) and revived by Geschwind (123), posited that two areas are vital for language: (1) Broca’s area in the left frontal opercular region, which encodes phonology for expression, and (2) Wernicke’s area in the left posterior superior temporal gyrus, which associates heard speech with meanings. Finally, a white matter bundle that connects these areas, the arcuate fasciculus, is incorporated by this model to indicate how Wernicke’s area may control speech expression (125).

This "disconnection" model was popular because it localized the perisylvian aphasia subtypes; thus, Broca aphasia, following left basolateral frontal injury, primarily affects expression and minimally affects comprehension. Wernicke aphasia, following left lateral temporal lesion, shows the opposite pattern. Global aphasia typically follows larger lesions that affect both sites. Finally, conduction aphasia, which frequently follows lesion to an intermediate site, may be seen to disrupt the coordination between Broca's and Wernicke's areas via the arcuate fasciculus, without severely affecting either expression or comprehension. That the individual may continue conversing is explained by postulating a vaguely-described alternate route between Wernicke's and Broca's areas that bypasses the arcuate fasciculus. The alternate route arouses semantic associations from words that are heard, but does not enable accurate speech repetition. Consequently, the conduction aphasic may make semantic paraphasic errors on formal repetition testing (123). Structural neuroimaging support for the primacy of the arcuate fasciculus has been noted, in that the extent of stroke lesion involvement with the arcuate fasciculus does more to predict damaged language functions than does lesion to other white matter tracts (212).

However, this model has many deficiencies (67). It fails to explain the preponderantly phonemic paraphasic errors of conduction aphasia (50; 83). The bidirectional and multisynaptic relays between Broca's and Wernicke's areas are overlooked (50; 221). With rare exceptions (327; 12), subcortical lesion of the left arcuate fasciculus does not lead to conduction aphasia (304; 12; 232). Similarly, Rao has observed that so-called classic "disconnection" disorders are not reported in multiple sclerosis [with the rare exception of Arnett and colleagues’ report (12)], which commonly targets white matter tracts (271). Most often, conduction aphasia, following stroke, involves cortical as well as subcortical injury (84; 269), as do other perisylvian aphasias (85). Although conduction aphasia primarily follows cortical lesion between Wernicke's and Broca's areas, the most severe and enduring repetition disturbance follows damage to Wernicke's area itself, even among individuals with improving speech comprehension (294); hence, the model does not explain the differential impairment of repetition according to lesion site. During silent visual naming (but not line orientation judgment), simultaneous electrocortical activation in anterior and posterior perisylvian sites occurs (114; 251), contrary to a serial processing model of language. Grammatical errors are not explained (124). Metabolic research in healthy adults implicates multiple left hemispheric areas in addition to Broca’s and Wernicke’s areas in aspects of propositional speech, including the rostral left ventrolateral cortex, the area immediately anterior to the supplementary motor area, and the right cerebellum (see below) (41).

Partly in response to these observations, research has explored whether additional white matter pathways may be involved with language processing. Converging opinion postulates that at least a second demonstrable tract, separate from the arcuate fasciculus, is equally important for language (341). It links the left frontal and temporal cortical areas ventrally and occupies the extreme capsule. Functional imaging research in healthy subjects suggests that this inferior pathway is activated primarily during speech comprehension (288). In contrast, the superior pathway (including the arcuate fasciculus) is believed to be primarily utilized during sound-to-speech associations, ie, speech repetition. However, the clinical relevance of the identification of these separate pathways has yet to be elucidated. Preliminary studies suggest that inhibitory stimulation to these various pathways may disrupt distinct linguistic functions (eg, resulting in either semantic paraphasia or phonological paraphasia), but surgery that disrupts these fiber pathways does not provoke a sustained language disorder (211).

In addition, the serial processing or disconnection model is becoming supplanted by an alternate concept. This network, or parallel distributed processing model (03), posits that the brain has regionally specialized functions, but that particular cognitive operations emerge from reciprocal neuronal communication among these diverse areas; thus, restricted lesion does not abolish a regionally specific function, but instead impairs it. Compensated function (recovery) emerges from the enhanced activation of, or reorganized connections among, surviving brain regions. The primacy of Broca area has become demoted in recent research, which shows that persisting impaired naming is more affected by insular white matter damage rather than direct injury to Broca area (120). Nonetheless, the model is incomplete in that it does not well address grammatical disturbances or phonemic paraphasias.

Advanced imaging studies have indicated changes in cerebral cortical activation in the perisylvian aphasias. For example, in primary progressive aphasia, a degenerative disorder that mainly damages the left perisylvian cortex, functional neuroimaging demonstrates activation of cortical areas adjacent to perisylvian cortex during linguistic tasks (310). Such responses may either reflect compensatory activation of less damaged cortex during linguistic tasks, or cortical disinhibition secondary to tissue injury. Right hemispheric hypertrophy in cortical areas that are homologous to left hemispheric language areas appears in response to left hemispheric infarction and aphasia, suggesting that the right hemisphere assists with aphasia recovery following left hemisphere infarction (351).

Extra-perisylvian contributions. Because the perisylvian structures engaged in language processing are anatomically and, thus, also physiologically linked to other brain areas, it is not surprising that research findings indicate that the integrity of such “remote” areas may bear upon linguistic functions. The right cerebellum is metabolically linked with the left frontal cortex on language-related tasks, as demonstrated on positron emission tomography studies of word generation and speech discrimination (261; 270; 172; 220). It is, therefore, not surprising that left frontal injury resulting in nonfluent aphasia is associated with contralateral cerebellar hypometabolism, and that this is less common in fluent aphasia (230). Whether contralateral cerebellar metabolic dysfunction following left frontal lesion contributes to aphasia is unknown. However, several case reports have associated acute right cerebellar lesion with left cerebral hypometabolism and agrammatic speech (305; 213; 355). Nonetheless, these reports did not consistently indicate selective perisylvian involvement when the patients were symptomatic. Silveri and colleagues suggest that right cerebellar lesion may be associated with agrammatic speech due to a working memory impairment that disrupts the modification of words according to their grammatical roles, leaving behind their root forms. Zettin and colleagues observed that grammatical production improved when their patient recited well-learned speeches from memory. Possibly, his increased concentration during this "nonpropositional" task benefited the memory impairment postulated by Silveri and colleagues; however, Zettin and colleagues suggest that cerebellar lesion directly disrupts the "procedural knowledge" of grammar, and not merely working memory. It is notable that none of these studies reported marked comprehension disturbances in their patients, which suggests that right cerebellar dysfunction chiefly impairs grammatical production when aphasic disturbances appear.

Some studies also indicate a hippocampal contribution to language. Successful rehabilitation for acutely impaired naming has been associated with bilateral activation of the hippocampal formations (228). Months later, this activation was supplanted by activation in the right temporal lobe and, to a lesser extent, left hemisphere peri-infarct temporal areas. In agreement with this research, it has also been shown that the involvement of focal brain damage in aphasic patients to temporobasal areas, adjacent to the hippocampus, more greatly impairs recovery and response to rehabilitation than damage that spares these areas (133; 226). It may be of importance that visuospatial working memory has been correlated with the extent of linguistic recovery in aphasic patients following rehabilitation (298), thus, suggesting the involvement of memory or executive processes in an as yet undetermined manner with aphasic treatment outcomes.

In related research, lesions that involve the basal ganglia have been associated with reduced language recovery after aphasia rehabilitation whereas lesions that extend to more anterior brain areas are associated with improved language recovery (256).

A further observation is that the consequences of solitary, left-hemispheric infarction that causes chronic aphasia is associated with cerebellar atrophy (246). Moreover, the variation in cerebellar volume predicts concurrent aphasia severity. These observations imply that solitary infarction that causes aphasia may have widely distributed structural brain changes that in turn affect aphasia severity.

Epidemiology

Studies have consistently demonstrated that nonfluent aphasics tend to be younger than fluent aphasics (25; 89; 231; 52; 103). Such differences appear not only following stroke, but also traumatic brain injury (52) and neoplasm (231), which suggests that hemodynamic factors do not exclusively explain (if at all) these trends among stroke patients. However, posterior infarcts are reported to predominate in the elderly, whereas anterior infarcts have no age bias (103). It is also worth noting that traumatic brain injury and neoplasms are more often associated with fluent than nonfluent aphasia (25). The reasons for this are unknown. Perhaps, the greater amount of focal cerebral necrosis in stroke than in traumatic or neoplastic disease affects the capacity to recover fluency.

Acute mild traumatic brain injury has been found to have greater fractional anisotropy (degree of linear water molecule diffusion) in the right hemisphere’s arcuate fasciculus (major intrahemispheric white matter bundle specializing in language) than in healthy controls, whereas the left hemisphere version of the tract is not different from healthy controls (337). The reason for the selective effect on the right arcuate fasciculus structural integrity in traumatic brain injury is unknown. The integrity of the right arcuate fasciculus in healthy controls is negatively correlated with the performance on naming tests, which may favor left hemisphere specialization for language. The absence of such relationship in persons with traumatic brain injury may impede language function, specifically, naming.

Cerebral lateralization for language may progress during the lifespan (69), which may lead to more fluent than nonfluent aphasia later in life following brain injury. However, an alternate explanation is that nonfluent aphasia is more likely to involve larger stroke lesions (25), which are more likely to be fatal among the elderly. Therefore, elderly patients with large lesions that should produce nonfluent aphasia may not survive long enough to be included in epidemiologic aphasia surveys (25; 76). However, this does not explain why nonfluent aphasia is so much more common than fluent aphasia among young adults. An alternate, but not incompatible, possibility is that age-related fluency changes leave fluent aphasia more likely to appear among the elderly. This has been supported by observations of speech patterns in healthy adults of diverse ages. Elderly individuals are more "fluent" or verbose; therefore, this feature may increase fluent aphasia among the elderly (69).

Differential diagnosis

Confusing conditions

Aphasia must be distinguished from other disorders that disrupt spoken communication. Hearing impairment can be mistaken by the clinician for receptive aphasia (93). Often, hearing-impaired patients benefit simply from being spoken to loudly. Impaired speech comprehension with retained reading comprehension suggests disordered hearing, rather than aphasia; therefore, writing instructions to the patient who does not comprehend speech may help indicate whether a primarily auditory disorder is present. Nonetheless, one should also consider pure word deafness or auditory agnosia in such instances in which hearing reception is normal, but the patient cannot interpret sounds (215). Such individuals may nonetheless speak, read, and write normally. Formal audiometric examination is essential when diagnosing these latter disorders.

Speech dysarthria or aphemia is also often mistaken for aphasia. If the patient appears to slur his speech, yet produces the usual parts of speech and whole sentences without struggling to choose words and without producing unusual words, the patient likely does not have aphasia.

The examiner must carefully regard patients whose heavily accented speech is due to a regionally distinct dialect or whose first language is not the same as the examiner's. In such cases, the examiner should seek help from someone who is familiar with the patient's linguistic background.

Certain specific motoric speech disorders that do not involve oral weakness should be distinguished from aphasia. Stuttering is a noticeable impairment in speech production that causes involuntary repetition, prolongation, or pausing in sounds (02). Stuttering resembles the repeated syllable production in conduction aphasia, but differs in that the repetition in conduction aphasia appears to be an attempt to correct speech errors, whereas the repetition in stuttering is involuntary. Nonetheless, acquired stuttering frequently appears in aphasia (151). Furthermore, developmental stuttering is related to perisylvian aphasia in that both involve structural abnormalities in the perisylvian cortex (110). Apraxia of speech is a controversial disorder in which the success in producing specific speech sounds varies from time to time, even when repeating the same word (313). Other language functions are not impaired; however, individuals diagnosed with apraxia of speech often have aphasia as well, usually the Broca variety.

The absence of speech with intact comprehension and writing suggests either mutism from a cerebral deficit, or aphonia from a peripheral nerve or vocal cord deficit.

Aprosodia refers to impaired processing of speech intonation (variation in pitch or volume), rather than of syntax, phonology, or semantics; thus, some patients may be impaired conveying their emotional tone despite producing intended words correctly (expressive aprosodia), whereas others may be impaired interpreting another individual's mood through his or her speech (receptive aprosodia). Such emotional disturbances appear to be most pronounced following right hemispheric injury.

Psychotic disorders, particularly schizophrenia, may have unusual speech that superficially resembles the output of fluent aphasia; however, several features distinguish schizophrenia from fluent aphasia (121). Most schizophrenic patients have minimal comprehension disturbance. Aphasic patients often try to improve communication by restating garbled speech or by gesturing, whereas schizophrenic patients are more likely "autistic," less concerned with overcoming the examiner's misunderstanding. Schizophrenic patients are distinguished also by bizarre thought content, including fantastic imagery, persecutory content, or other delusional fixation, whereas aphasic patients perseverate characteristically on short phrases or words, but without emphasizing a strange theme. Schizophrenic patients are unlikely to have paraphasic errors.

Other abnormal speech productions should be distinguished from aphasia. These are covered in the classification by Mendez, which in most instances involve premeditated disruptions (227). These include oxylalia (rapid speech), glossolalia (speaking in an apparently unknown language, encountered most often in religious fervent possession), and coprolalia (explosive offensive speech), which is most often associated with Tourette syndrome, and although it can be repressed, it is accompanied with a sense of urgency.

Delirium may interfere with speech comprehension and, therefore, the two disturbances may be confused (277); however, delirium involves lack of sustained attention to all kinds of stimuli, not merely to speech. Disturbances of arousal (eg, sleepiness, intoxication, and coma or other "minimally aware" states) should not be mistaken for aphasia.

Associated or underlying disorders

A growing area of research evaluates the nonlingustic co-occurring deficits in aphasia. The vast majority of post-stroke aphasic patients score poorly on brief general cognitive assessments (88). It has been long recognized that the speech repetition deficit in conduction aphasia may result in part from impaired short-term memory (56). Nonverbal memory is impaired in a variety of aphasias (192). Nonverbal tests of reasoning (the Raven’s Colored Progressive Matrices) and attention shifting (the Wisconsin Card Sorting Test) predict language impairment in aphasia (18). Patients who score low on standard aphasia assessments are also more likely to be impaired with sustained attention on a test involving continuous movement performance, the Conner Test (198). Nonlinguistic functions improve in parallel with language recovery following rehabilitation for global aphasia (214). Aphasia is so much associated with general attention deficits that aphasia may be considered essentially as a language-specific attentional disorder, coupled with nonverbal attentional deficits. Moreover, the severity of attentional deficits in aphasic patients may predict the extent of linguistic recovery (335).

These results suggest that aphasia may involve diverse cognitive impairments, although further work is needed to exclude comprehension deficits as a confounder toward cognitive testing in aphasia.

Diagnostic workup

Patients with impaired communication that is suspected to be due to aphasia should be tested further to indicate their limitations to themselves, family, and clinicians. The typical aphasia exam assesses four basic functions: (1) conversation, (2) comprehension, (3) repetition, and (4) naming. Informal conversation can indicate agrammatism, impaired comprehension, anomia, fluency problems, and paraphasic disturbances. A good summary and convenient manner of testing is provided by Kertesz and Poole (178). Patients with greatly inhibited speech initiation can be successfully engaged in conversation by telephoning them, which appears to prompt an automatic behavioral response in a more familiar context (49). Comprehension should be assessed by having the patient reply with yes or no answers or appropriate gestures. The examination should include questions that definitely have no for an answer (eg, "Are you growing a horn?"), because some severely aphasic patients will agree almost with anything, perhaps to please the examiner. Comprehension also may be assessed by having the patient point to named objects within the room or on the body, but one must be careful that responses are not biased to one side due to unilateral neglect. Failure to perform a requested learned action (eg, "Wave good-bye") may also be due to apraxia, rather than aphasia; therefore, yes or no testing may be more reliable. Regardless of the task used to assess comprehension, the examiner must be careful not to visually cue the patient to the answer, such as through subtle head motion or glancing at a named object.

Repetition testing may start with single, one-syllable items and progress to multi-syllable words and sentences. Often, however, repetition testing can be simplified by asking the patient to repeat "No ifs, ands, or buts." This unusual phrase (which lacks concrete nouns and a verb) will usually stump perisylvian aphasics, unlike conventional short sentences. Anomia may be tested through having the patient name generally familiar items in the room such as a pen, watch, telephone, and so on. The examiner must first ascertain that vision is adequate for this task, such as by testing finger counting. Visually impaired patients may be tested through their palpating or listening to objects (eg, comb, crumpled paper, keys). When patients cannot name, their recognition of the object should be verified through having them describe either the object's function or where it may usually be found. Failure to succeed on such object description suggests the problem may actually be due to agnosia, rather than anomia.

From such evaluation, which may be conducted in 10 to 15 minutes, the examiner can provisionally identify the kind of aphasia. Aphasic patients without previous work-up should be referred to a speech pathologist for more thorough assessment.

In suspected acute onset aphasia without noticeable progression (eg, following stroke or traumatic brain injury), cerebral neuroimaging may help to confirm the impression by demonstrating a left hemispheric focal lesion, usually in perisylvian cortex or subcortex. Rarely, a solitary acute right perisylvian lesion may be found instead, which should not rule-out the diagnosis of aphasia. Care should be taken to insure that the neuroimaging study was performed at least 24 hours after aphasia onset if one wishes to localize the responsible injury. Electrodiagnostic study (eg, auditory evoked potentials) may help to determine whether profoundly aphasic individuals can interpret speech and may aid prognosis (75).

Key tests and procedures

Several formal and validated aphasia test batteries are commercially available and are administered primarily by speech-language pathologists. Among the leading assessments are the Boston Diagnostic Aphasia Examination (135), Western Aphasia Battery (177), Porch Index of Communicative Ability (265), and Aachener Aphasie Test (the last one is administered in German) (159). These formal assessments generally assess the main categories of language: conversation skills, naming, repetition of words and phrases given to the patient, and comprehending commands.

A limitation, little addressed in the literature, is that formal test batteries assess language skills under artificial settings (clinic, laboratory), in response to command by the tester, and do not assess real-world conditions and spontaneous conversation. Other research in the area of motor capabilities after stroke has demonstrated that there can be considerable divergence between performance under artificial settings compared to spontaneous activity in the real world (09). Consequently, formal testing by these batteries may not accurately reflect real-world speech after brain illness, which is more important to the patient. As an alternative assessment, the Verbal Activity Log, or VAL, has been developed (140). This is a self-reported assessment of the quality, and separately, the amount of speech in 12 everyday conditions (eg, answering the telephone, ordering a meal). The amount of speech is significantly correlated with the amount of real-world speech captured by a digital audio recorder worn by the patient. The quality rating was also correlated with the amount of speech as recorded by the audio recorder. Although the methods of the VAL have been made readily available, only one other research site has used this assessment thus far (168). An obvious limitation of the VAL is that comprehension impairment with the patient could prevent its administration.

Management

Aphasia management, in part, depends on identifying and treating the underlying disorder to prevent its progression or recurrence; therefore, workup should include cerebral imaging studies (usually CT or MRI) and, where appropriate, EEG, lumbar puncture, carotid artery and cardiologic evaluations, and surgical consultation. Referral to a speech pathologist is advisable to confirm the clinical impression of aphasia, determine the functional limitations for communication, and identify methods to treat or circumvent the communication disorder.

The nonspecialist can follow some simple measures when treating comprehension failure. Patients with acutely impaired comprehension may become agitated, which may be compounded when clinical staff members fail to recognize the disorder and react negatively. Such patients may respond well to soothing emotional tones, a calm attitude, and being addressed slowly (205; 339; 277). Because abrupt changes in topic impede comprehension among healthy subjects (55), maintaining topic consistency is even more important when addressing aphasics. Communicating through gesture and speech combined may benefit patients with severe comprehension impairment (36). When speaking with colleagues or family members in the same room, clinicians should not ignore severely aphasic patients or believe that they cannot comprehend. Comprehension may be better than is clinically suspected (07; 100), and even severely aphasic patients may comprehend emotional meaning from speech (21); therefore, courteous attitude toward such patients should always be maintained. Conversely, clinicians should not mistake a severely aphasic patient's nodding along with everything that is said for retained comprehension. Family members may assist the clinician's communication with the patient through their knowledge of the patient's preferences and expressions and through the patient's familiarity with their faces and intonation. Aphasic patients often respond appropriately to facial expressions, vocal intonation, drawings, and gestures, although comprehension of the latter, nonetheless, may parallel verbal comprehension ability (119). A board printed with symbols or drawings of items that are commonly used in the hospital, such as a drinking cup or a pillow, and simple words (eg, yes, no, hurt) may be used by the patient to point to vital needs, provided that the clinician can establish that the patient can use these representations reliably.

Written expression and comprehension among aphasic patients has been reported to benefit from intensive practice (65). An application of writing training has been in the incorporation of training to improve texting, which was shown to benefit text communication in one case report of severe aphasia (28).

Speech pathologists may utilize a wide variety of techniques, based on identifying the foremost deficits and then training patients to utilize progressively more complex responses or stimuli. Preliminary research suggests that the kind or severity of aphasia may determine optimal forms of speech therapy. Thus, fluent aphasia may respond better to semantic-oriented therapy, whereas the most severe aphasia may respond better to phonological therapy (191). One popular treatment, melodic intonation therapy (MIT), capitalizes on the relatively spared ability of some aphasic individuals to sing (249). This intensive and extended treatment has been shown to result in increased “fiber count” of the undamaged, right arcuate fasciculus as measured by diffusion tensor imaging among individuals with Broca aphasia (290). Nonetheless, a preliminary randomized clinical trial of MIT versus waiting list showed poor retention of speech improvement from the former (331).

An alternate approach that has gained acceptance involves training speech production while immersing the patient in a set of behavioral procedures that center on operant conditioning (ie, rewarding attaining specified goals). The initial method was termed Constraint-Induced Aphasia therapy (CIAT), or Constraint-Induced Language Therapy (CILT). CIAT was originated in 2001 by Pulvermüller and colleagues as a derivation of Constraint-Induced Movement therapy (CI therapy) to improve chronically, behaviorally repressed use of a partly paralyzed arm after brain disease (268). To date, trials of CIAT have not had a clear superiority over other forms of treatment (356), although there have been, by now, more than 44 clinical studies of CIAT and its variants. In response, a modification of CIAT (termed CIAT II) was tested in a pilot study, which included behavioral techniques to extend gains from the clinic to real-world conditions (167). The preliminary results showed larger effect sizes on speech outcomes than in other trials in CIAT. Larger trials will be needed to evaluate the consistency of these gains. Further support for trials of CIAT, or other forms of speech therapy for aphasia, would include measuring real-world spontaneous speech in addition to speech under artificial laboratory or clinic environments. A short, patient-reported instrument termed the “Verbal Activity Log” (VAL) was validated to measure the quantity of real-world speech when compared against simultaneous objective audio digital recordings (140).

All together, these findings suggest that beneficial speech therapy for aphasia may be associated with enduring structural neuroplastic changes in the brain. Unfortunately, in conventional aphasia therapy, no standardized approach to aphasia therapy is used, which is partially because there have been few carefully controlled studies that have evaluated specific treatment variables, such as duration of therapy and time since illness onset. In addition, patient characteristics that might influence these outcomes are undetermined (158). Interindividual consistency in therapists' applications of specific techniques has also been difficult to control (10).

Aphasia following stroke is strongly associated with nonverbal disability, for unclear reasons (109; 129). Hence, finding aphasia should prompt the evaluation of competency for self-care skills. However, many case reports indicate that patients with profound aphasia may nonetheless be independent on nonverbal activities of daily living and may even resume meaningful employment (236; 146; 24; 345; 272; 118; 278; 27; 147).

Several unconventional therapeutic approaches have been evaluated for treating aphasia in small-scale studies and offer some promise for further rehabilitation advances.

Medications. A plethora of medications has been used to treat aphasia, with few replications of treatment results. Despite initially promising findings, nonfluent aphasic patients following stroke or traumatic brain injury have not been found to benefit from the dopamine agonist bromocriptine in placebo-controlled trials (138; 280). However, levodopa has been found to benefit fluency and repetition in aphasic patients, particularly following anterior cerebral lesions (299). A second study, however, found no benefit to levodopa (199). Low-dose amphetamine paired with speech therapy has benefited aphasia in a preliminary report of a double-blind, placebo-controlled study (336). A similar outcome was shown in a single case study (312); the authors suggested the outcome reflected improvement of apathy rather than a direct linguistic benefit. A few reports have indicated improved language functions following drugs that augment acetylcholine activity (209; 165; 320; 34; 20). Other studies have identified benefits to anomia following either selective serotonin reuptake inhibitors (SSRIs) (319), propranolol (37), or dextroamphetamine (349). Preliminary reports of treatment with piracetam have indicated improved aphasia recovery following stroke, perhaps due to protective effects on cerebral metabolism (92; 160; 189; 180). Other reports have found beneficial effects following vasopressin (30) or donepezil (33; 71). Preliminary findings indicate that a gamma aminobutyric acid (GABA) agonist (midazolam) reinstates recovered aphasia following stroke (196), thus, suggesting that GABA antagonists may benefit the recovery of aphasia. In contrast, an interesting case report serendipitously found that zolpidem, another GABA agonist, reduced the involuntary production of stereotyped syllables in a chronically aphasic patient (74).

Gestural communication. In certain cases visual comprehension may exceed auditory comprehension ability; therefore, gestural communication (eg, through formal sign language) may be a practical substitute for vocal communication (184). Limb apraxia, which frequently occurs with perisylvian aphasia, may impair gestural communication. This may improve with training that pairs gesture with spoken words, particularly for verbs (274). In addition, pointing with the paretic right arm toward objects to be named may elicit more successful responses than pointing with the intact left arm, among patients with various nonfluent aphasias, but not in fluent aphasia (143).

Phonomotor therapy. Phonomotor therapy is an advanced treatment for impaired word retrieval (anomia) in aphasic patients (175). The treatment involves having the patient observe the therapist’s mouth movements for target words, for the patient to recreate these over several hours per week and for 6 to 7 weeks. The goal is to produce words that are phonologically (motorically, pertaining to mouth parts) similar to the target words. A randomized controlled trial of phonomotor therapy observed no generalization to untrained words (174), and so it appears that this approach is not likely to be helpful.

Orthoses. Physical tools may improve expression, particularly among global aphasics. One approach is to attach the global aphasic patient's paretic right arm to a device that grips a pen and permits it to smoothly glide over a sheet of paper, through the patient's voluntary propulsion at the shoulder. Surprisingly, this may allow crudely executed writing that is nonetheless more accurate on naming and dictation tasks than that accomplished by the nonparetic arm (53; 201; 115). This approach’s lack of popularity may be, in part, due to lack of consensus on the basis of this behavior (207). An alternate approach is to provide the nonfluent aphasic patient with a computer that can be used to communicate desires through manipulating iconic symbols on the screen with a mouse (342). Severely aphasic patients may require extensive training, but preliminary findings indicate that such patients may improve in their comprehension abilities as well.

Transcranial magnetic or direct electrical current stimulation. Preliminary study of transcranial magnetic stimulation (TMS) to the left temporal cortex in control subjects has shown reduced response times for naming tasks (240). Such treatment when applied to the right frontal lobe was related to improved picture or object naming in several forms of chronic aphasia (Broca, anomic, global), with sustained benefits in most subjects (241; 142; 242). Inhibitory transcranial magnetic stimulation (TMS) directed to the right hemisphere Broca area equivalent has been associated with aphasia improvement compared to stimulation to the cranial vertex (340). A related treatment, transcranial direct current stimulation (tDCS), has been found to promote learning new words for familiar objects in healthy subjects (107). Initial studies have found tDCS to benefit naming among aphasic patients (237; 15; 99; 106; 108; 169). However, transcranial magnetic stimulation has been shown to have improved naming when compared to tDCS to the right hemisphere, but only in patients with structurally intact Broca area (357). Direct epidural electrical stimulation has been shown to improve the treatment effects of conventional speech therapy for aphasia (72). Therefore, transcranial electromagnetic or direct electrical cortical stimulation approaches suggest some promise for improving expressive deficits in neurologic disease.

Outcomes

Extended speech therapy is generally associated with improved communication (158). However, it is unclear whether any particular standard technique is successful because some studies have found that nonspecific positive interaction is as effective as formal speech therapy (224; 87; 145). Another study found no differences in outcome between patients receiving treatment and those who did not (206). However, these studies assessed small sample sizes or used limited speech therapy contact for comparison, so it is unclear whether speech therapy is any better than positive reinforcement. An overview of treatment studies indicated that aphasia following stroke is more likely to improve with more intensive speech therapy (39). However, this finding was contradicted by a randomized clinical trial that found no difference in subacute aphasia recovery between speech therapy for two hours per week versus five hours per week (16). Research has suggested that tDCS maintained language improvements in patients with chronic aphasia for as long as six months in follow-up, although the effect size on real-world communication was modest (225).