In 1861 French neurologist Dr. Paul Broca provided perhaps the most well-known initial description of a language impairment subsequent to focal brain damage (Broca, 1861). In a now famous case, Broca described a patient who had lost the ability to produce speech, with the exception of a single word (“tan”), yet demonstrated spared comprehension. Upon the patient’s death and subsequent autopsy, Broca discovered damage to the left inferior frontal region of this patient’s brain (Figure 10-1, BA 44-45), and he therefore attributed speech output abilities to this circumscribed neural region. In this way, “Broca’s aphasia” came to be known as primarily a speech production disorder following damage to left inferior frontal regions of the brain.

Figure 10-1 Left hemisphere of brain with Brodmann area labels overlaid. German neurologist Korbinian Brodmann (1909) provided a roadmap for the organization of the human brain by painstakingly mapping 52 cortical regions. Each area, now known as *Brodmann areas* (BA), is a region of the cerebral cortex that is defined based on its cytoarchitecture, or organization of cells. Historically, regions which have been implicated in language processing are BA 44, 45, and 22, although a number of neural regions have now been demonstrated to contribute to language performance.

Ten years later, German neurologist Dr. Karl Wernicke was also investigating the relationship between localized brain damage and language impairments (Wernicke, 1874). He noted that language deficits could occur following damage to parts of the brain other than the left inferior frontal region, and that symptoms differed depending on the brain regions involved. Specifically, Wernicke posited a link between the left superior temporal gyrus and language comprehension (see Figure 10-1), after observing that patients with damage to this region demonstrated comprehension impairments. From this, “Wernicke’s aphasia” was characterized as a disorder of language comprehension following damage to the left superior temporal gyrus. It was the work of Broca and Wernicke that spurred early investigations of brain injury and its effect on language (among other behaviors).

Two important observations should be noted regarding “aphasia” research during this time.³ First, scientists were operating under the assumption that particular language functions (e.g., production, auditory comprehension) were controlled by specific neural regions, and thus assumed that damage to that region resulted in a specific language deficit. While this characterization of brain–behavior relationships is intuitively appealing, later in this chapter we will discuss how a good deal of research has since shown this line of thought to be inadequate. Language is much more complex than a simple “input–output” system.

During the early era of aphasia research, the field of aphasiology was descriptive—language disorders were characterized based on direct observable behaviors (i.e., the patient’s overt symptoms). However, several years after Broca and Wernicke described patterns of language impairment in their patients, the Wernicke-Lichtheim model was introduced (Lichtheim, 1885), which marked the first attempt to predict patterns of aphasia following stroke. This model, shown in Box 10-2, anticipated aphasia subtypes, or groups, based on symptoms and site of neural trauma. Symptoms were cast in terms of language activities, like speaking and listening; “lesions” (injury) within the core areas and the connections between them were used to predict patterns of language impairment in aphasia. For example, damage to the motor center would yield a Broca’s type of aphasia, while a lesion to the auditory center would yield a Wernicke’s type of aphasia. A lesion to the auditory-motor pathway would yield a disconnection syndrome, such as conduction aphasia, characterized by the inability to repeat what is heard. Hence, this model ushered in an era of aphasiology that could be characterized as “connectionism.⁴” During this era, connectionism assumed that higher mental functions were dependent on the connections between different centers in the cortex (Ahlsén, 2006). A more modern proponent of connectionism, Norman Geschwind, advanced the theory that disconnections between neural regions could lead to language disturbances (Geschwind, 1965). Connectionism remained an important concept in aphasiology for many years (and variants of this work remain active topics of research).

Box 10-2 Adaptation of the Wernicke-Lechtheim Model of Language Disturbance in Aphasia

Predicted Deficit	Pathway
Auditory comprehension	Auditory input	→	Auditory center	→	Concept center
Voluntary speech	Concept center	→	Motor center	→	Motor output
Repetition	Auditory input	→	Auditory center	→	Motor center	→	Motor output

The contemporary era of aphasia research (and the focus of this chapter) was ushered in by Caramazza and Zurif (1976) and others, who examined the dissociation between different levels of language processing deficits in aphasia (see Accounts of Language Deficits in Aphasia later in this chapter). Indeed, current research aims to uncover the processing deficits in aphasia that may account for observable language deficits. Before we discuss this process-oriented research, we first introduce and briefly describe the major types/syndromes of aphasia, to set the stage for a discussion of processing differences between these syndromes.

Types of aphasia

In this section we present some of the major classifications (or subtypes) of aphasia. Note here that we are describing the prototype of each of the main subtypes of aphasia. “Pure” cases of any given type of aphasia are seldom observed.

Broca’s aphasia is typically characterized by halting, nonfluent speech. Grammatical function words (such as “is,” “and,” “the”) and inflections (for tense, agreement, number, gender), are often omitted in language production (spoken or written), with mainly root content words being produced. Someone with Broca’s aphasia may say “boy girl fall” to mean “The boy and the girl are falling” (Figure 10-2, A, B). Language production is therefore described as telegraphic or agrammatic (“without grammar”). In the 1970s, it was discovered that this agrammatic component of Broca’s aphasia also extends to comprehension. The failure to detect comprehension deficits in Broca’s patients was likely due to the fact that comprehension of single words and simple sentences (mostly) appears relatively intact. We now know that language impairment in Broca’s aphasics extends to comprehension of more complex sentence constructions such as noncanonical structures (i.e., passives and object-relatives, see Chapter 6). This impairment for noncanonical sentences was investigated by Caramazza and Zurif (1976), who noted that Broca’s patients demonstrated poor comprehension for sentences containing reversible noun phrases (e.g., “The boy was chased by the girl,” where either of the nouns boy or girl can perform the action chase). This was in stark contrast to the group’s ability to understand noncanonical sentences that did not contain reversible constituents (e.g., in “The ice cream was eaten by the boy,” only the animate noun boy can perform the action eat, thus reducing the likelihood that patients would believe ice cream to be the subject of the sentence). This finding of Broca’s patients’ difficulty in understanding noncanonical sentence structures led to the term “overarching agrammatism,” which refers to the fact that the grammatical constituents of language are not employed in either the production or comprehension of sentences (this will be discussed in more detail shortly). In addition to agrammatism, repetition abilities may also be impaired in this group (Table 10-1).

Table 10-1

Types of Aphasia and Associated Symptomotology

Aphasia Types	Production	Comprehension	Repetition	Naming
*Nonfluent*
Broca’s	Halting, agrammatic	Impaired for noncanonical syntax	Poor	Poor
Transcortical motor	Halting, agrammatic	Impaired for noncanonical syntax	Intact	Poor
Global	Severely impaired	Severely impaired	Poor	Poor
*Fluent*
Anomia	Fluent, with word-finding problems, circumlocutions	Intact	Intact	Moderate-poor
Wernicke’s	Fluent, facile, paraphasias	Poor	Poor	Poor
Transcortical sensory	Fluent, facile, paraphasias	Poor	Intact	Poor
Conduction	Fluent, facile	Intact	Poor	Intact

Broca’s aphasia, and a subtype known as agrammatic aphasia, commonly occur following damage to the left inferior frontal regions (Figure 10-3).⁵ However, there have been reports of individuals with Broca’s-like behaviors who have sustained damage to other, subcortical, regions (Alexander, Naeser, & Palumbo, 1987; Fridriksson, Bonilha, & Rorden, 2007), and inconsistencies of this sort make it difficult to precisely characterize brain–behavior relationships.

Figure 10-3 An illustration of typical sites of neural injury resulting in aphasia. This illustration demonstrates the typical 1:1 lesions associated with these syndromes. It is important to not confuse the correlation of structure with that of function. There is considerable variability in lesion sites with respect to damage to cortical and subcortical tissue.

In contrast to the halting, agrammatic speech observed in Broca’s or transcortical motor aphasics, a patient with Wernicke’s aphasia has facile language production and speech that contains grammatical structure. In this type of fluent aphasia, the “preserved” speech fluency originally led researchers to regard Wernicke’s aphasia solely as an impairment of comprehension. Research has since shown, however, that language production in this syndrome is also impaired and typically contains many instances of paraphasias, as exemplified in Table 10-2.

Table 10-2

Examples of Paraphasias

Paraphasia Type	Description	Example
Phonemic (Literal)	Substitution of one phoneme for another within a single word	/pun/ for /spun/ /tevilision/ for /television/
Neologistic (Jargon, Gibberish)	Substitution of word sounds phonetically and semantically unrelated to the target words	/glick/ for /pencil/
Semantic (Verbal)	Substitution of one word for another, but not always semantically related	/flew/ for /soared/

Paraphasias in Wernicke’s aphasia may include jargon or neologistic paraphasias (among others), which are utterances that sound like words of the speaker’s language, but are in fact not real words. For instance, a patient with Wernicke’s aphasia may produce a neologism like “glick,” which is not a true word but conforms to phonological rules of the language (Figure 10-4).

Figure 10-4 A, Speech sample from an individual with Wernicke’s aphasia, retelling the Cinderella story. B, An individual with Wernicke’s aphasia’s written description of The Cookie Theft picture. [A is Courtesy of the Cognitive Neuroscience Laboratory, SDSU. B from Goodglass, H., & Kaplan, E. (1972). *Boston diagnostic aphasia examination*. Philadelphia: Lea & Febiger.]

Neologisms are suggestive of either impaired word finding or semantic processing in Wernicke’s aphasia. As illustrated in Figure 10-3, damage to the posterior temporal regions is most often implicated in this type of aphasia.

Conduction aphasia was originally described by Wernicke. It is classically defined as a disorder of impaired repetition of auditory material. More recent research has indicated that word-finding is impaired in the disorder, and that individuals with conduction aphasia often produce phonemic paraphasias in ongoing speech (Baldo, 2008; Fridriksson, 2010). Injury to the supramarginal gyrus (see Figure 10-1, BA 40) and arcuate fasciculus is commonly associated with conduction aphasia (see Figure 10-3).

Global aphasia is the most severe of the aphasia types, in which both language comprehension and production are profoundly impacted, rendering the individual severely impaired. Brain lesions typically span across the frontal and temporal lobes. By contrast, anomic aphasia, or “Anomia,” is perhaps the mildest form of the disorder and is characterized by problems recalling words or names.⁵ Individuals with this impairment frequently use circumlocutions (speaking in a roundabout way) to express words that they cannot recall or produce. Persons with anomic aphasia may not show other obvious signs of impairment and may in fact produce relatively fluent speech. Comprehension may be mildly impacted in this group with repetition most often intact. Lesion localization in this group is the least reliable of all the aphasias. Most often, there is injury affecting temporal and parietal regions (Fridriksson, 2010).

Some words of caution: In any discussion of language disorders resulting from neural trauma, it is important to bear in mind that diagnoses and classifications (type and severity of aphasia) are based primarily on a patient’s symptomology, thus being descriptive in nature. The descriptive nature of this process, relying on overt responses during assessment, can result in overgeneralizing a patient’s ability or inability. This can cause difficulty in the interpretation of patient performance on language assessments as any one assessment may not capture the entirety of a language disorder. As an example, consider a situation in which a researcher or clinician wishes to measure a patient’s auditory language comprehension abilities and uses a task known as sentence-picture matching. Individuals are presented with a sentence and two (or more) pictures and are asked to choose the picture that matches the sentence. A patient with an underlying sentence comprehension deficit may be able to use a strategy rather than his/her linguistic knowledge to correctly perform the task, and thus this task may underestimate the full magnitude of a patient’s language comprehension deficit. As an illustration, recall the earlier discussions of overarching agrammatism; prior to Caramazza and Zurif’s work on sentence comprehension in reversible sentence constructions, it was thought that Broca’s patients maintained intact language comprehension, thus resulting in claims that these individuals had a primary production problem with intact comprehension. We now know that these claims were inaccurate and that many comprehension assessment results were based on the ability of Broca’s patients to use world knowledge to figure out the meaning of certain sentences (e.g., ice cream cannot eat boys).

Another note of caution: research on the neural underpinnings of language is still in its infancy. A great deal of research has shown that although a particular area of the brain may be implicated in a specific language function (either through patient studies or in functional neuroimaging experiments), one cannot assume that a particular brain region is solely responsible for the process under investigation (i.e., that a particular language function lives in a specific brain region). A damaged brain region that is implicated in language processing may have only been an important part of a network of different brain regions that were critical to that particular level of processing. Likewise, if experiments demonstrate a link between normal operation of a specific language process and a region of the brain, this does not mean that this area exclusively serves that process, to the exclusion of other cognitive processes. Thus, localization of function is not as simple as assuming a direct 1:1 relationship between specific brain damage patterns and subsequent language impairment.

We now turn to a discussion of language processing in aphasia and begin with a discussion of the different techniques used to measure behavior.

Until the 1970s, it was standard practice to use one methodological approach to examine language ability in patients with language disorders. Typically, this approach took the form of an untimed meta-linguistic task. Such a task measures language processing after the event of interest (e.g., after an entire sentence has been heard). These types of tasks are considered “off-line” measures and reflect what listeners ultimately understand a sentence to mean. They allow individuals to consciously reflect on the problem and use all the resources available to them at the time to determine the final meaning of a sentence. Examples from aphasia research include sentence-picture matching, grammaticality judgment, and paraphrasing (e.g., Caramazza & Zurif, 1976; Friedmann, 2006; Linebarger, Schwartz, & Saffran, 1983) (see Chapter 6, Box 6-2, for additional discussion).

While off-line methodologies are informative in providing information as to how individuals ultimately understand or interpret language, these tasks do not allow for an examination of the moment-by-moment operations of sentence processing as they are occurring in real-time. In an off-line task, if an individual demonstrates poor sentence comprehension performance, it is unclear why the individual could not accurately perform the task. For instance, consider a sentence-picture matching task in which an individual with Broca’s aphasia demonstrates poor performance with noncanonical sentences. A number of factors could contribute to this end result: the individual might have an underlying lexical deficit (knowledge of word meanings); an underlying structure building deficit (inability to link the incoming words to build structural relationships among them); or an inability to retain the sentence and match it to the pictures (working memory deficit). Thus, while informative, off-line methodologies alone do not allow for precise characterization of the underlying processes that contribute to language impairments. Additionally, the task demands of off-line methodologies may draw the participant’s attention to the linguistic material under investigation, and in doing so alter the nature of the comprehension process. From this point on, we limit our discussion of off-line tasks and focus primarily on evidence garnered from on-line experiments.

The study of language processing via on-line methodologies

To understand the complete details of the comprehension process, language scientists require methods that can capture the moment-by-moment unfolding of comprehension. “On-line” tasks do just this, and allow researchers to observe language processing both while it is occurring and without conscious reflection on the part of the participant. With on-line experimental techniques and the right kinds of experimental designs, researchers can examine the individual stages or components of language processing, and in populations with language disorders, investigate which of these stages may be impaired. In keeping with the goals of this chapter, we briefly summarize on-line methodologies that have been (and are currently) used to study auditory sentence comprehension in Table 10-3 (see also Chapter 6, Box 6-1).

Table 10-3

Methodologies Commonly Used in the Study of Language Processing in Aphasia

Methodology	Elements of Language Processing Studied	Strengths	Limitations
Event-related potentials	• Various levels of language processing: phonological, lexical-semantic, syntactic, discourse-level	• Good temporal resolution • No secondary task required • Not limited to single data point per stimulus	• Only indicates the signatures of language processing, not the processes themselves • Concern about participant attention if no secondary task used
Eye-tracking	• Lexical integration • Syntactic processing • Discourse-level processing	• Good temporal resolution • No secondary task required • Not limited to single data point per stimulus	• Only indicates the signatures of language processing, not the processes themselves • Concern about participant attention if no secondary task used • Possible use of strategy, may alter comprehension processes
Cross-modal priming	• Lexical access • Syntactic processing	• Indicates precisely which elements of the sentence are processed (“activated”) at point of interest	• Only one data point per stimulus item • Dual-task demand may be difficult for some patients

Taken together, these on-line methodologies can provide complementary evidence regarding the nature of language processing. These methods have been critical in the development of accounts of language deficits in aphasia.

Here, the verb walk is inflected for tense (e.g., present or past), and must also agree with the person, gender or number of the subject (e.g., the boy/boys). Now consider Figure 10-5, which represents a linguistic tree structure for such a phrase, and note that Agreement (Agr) and Tense (T) are represented by separate nodes (see Chapter 6 for a discussion of tree structure representations; example from Friedmann, 2006; Pollock, 1989; additional reference from Grodzinsky, 2000a). Here, the important element to note is that Tense is represented at a higher node than Agreement. Interestingly, a good number of cross-linguistic studies (many in which tense and agreement are more richly inflected than in English), have demonstrated that individuals with Broca’s aphasia show dissociation in the production of Agreement and Tense. For example, given the following prompt in (3), participants should produce a verb that is inflected for tense and must agree with the subject (past tense, third-person feminine singular, e.g., “jumped”; from Friedmann & Grodzinsky, 1997):