There are many types of language tasks utilized in mapping procedures. Before discussing specific tasks, some general principles of the language mapping procedure are presented. In terms of item selection, regardless of the modality being tested, the stimulus items need to be chosen such that the patient can respond quickly and without any hesitation. For each patient, several items may need to be screened, and difficult stimuli for that patient should be eliminated. Therefore, for every modality assessed during the procedure, clinicians will need a somewhat large pool of items as each patient may not find them all equally easy. Typically a set of eight to ten items for each modality is utilized in pseudorandom rotation during the procedure. Practice effects are not an issue when eloquent language cortex is temporarily rendered nonfunctional via electrical stimulation; therefore, it is not problematic to use the same set of stimuli repeatedly throughout the procedure.

It is important to avoid exclusively using yes/no questions, to limit the potentially misleading effects of patient guessing. It is not uncommon for patients to try to be overly compliant with this “test.” They are, with good reason, highly invested in the process and often try their best to “answer correctly” to, from their perspective, facilitate the process. As a result, they may try to guess at answers to “get it right and be a good patient.” Even if yes/no response formats are avoided, a patient may try to “guess” the appropriate response from the pool of eight to ten possibilities that they have become very familiar with over the course of the procedure. This can be avoided by counseling the patient in advance that they should not guess as this will only confound the results.

Timing of item presentation and response must be precisely coordinated. Specifically, items and responses must be provided during the stimulation epoch. For example, the clinician should not begin providing an auditory comprehension item prior to stimulation onset, and the response must be given before stimulation is discontinued for the site to be considered “cleared” for that language function. If the patient responds after discontinuation of the stimulation or not at all, then, provided other causes for response failure are ruled out (e.g., ADs or motor responses), the site is considered a “hit” for that function. Thus, the clinician who is controlling the stimulation and the clinician who is performing the behavioral testing need to be working together to ensure a reliable and accurate mapping result.

The necessary inter-trial interval will vary from patient to patient. While Devinsky, Perrine, Llinas, Luciano, and Dogali (1993) recommend 20 s between trials, some patients will be able to proceed at a faster pace without any decrease in their language quality or response reliability. However, patients with particularly irritable cortex resulting in frequent ADs (and perhaps seizures), those who are particularly fatiguable, individuals who are lower functioning cognitively, or those who are evidencing a great many “hits” during the mapping process may require longer inter-trial intervals to avoid overtaxing their ability to respond. This balance between expeditious completion of testing and avoiding difficulties from repeated, frequent stimulation requires ongoing assessment of both the clinical performance and electrographic responses.

The criteria for a pass (an area of cortex is “cleared,” i.e., it does not subserve that language function) or fail (i.e., a “hit”) vary across surgical centers. We utilize and recommend a procedure wherein any missed items clearly reflect pathognomonic performance (i.e., a “hit”) as all items utilized were answered correctly with 100 % accuracy at baseline testing. If the patient reliably misses a second item of the same type (e.g., two pictures are not named during separate stimulation trials), then that site is considered to be eloquent for that function. It is recommended that at least two trials of each task at an electrode pair are answered correctly to “clear” a cortical site. This serves to clarify the reliability of the response. If stimulation is applied adjacent to a language zone, then testing may result in intermittent correct and incorrect responses. Such a phenomenon might be classified as a partial hit and indicate that the underlying brain region may subserve language functioning but that a postoperative language deficit might not occur if it were removed.

If the patient appeared to lose focus or the results were unclear from the initial stimulus trials, additional items can be administered. Prior to continuing with stimulation testing, baseline testing (i.e., testing without stimulation) should be performed again. Thereafter, presentation trials should proceed with intermittent stimulation during some of the item trials to look for disruption of a correct response pattern. We will often utilize this intermittent stimulation paradigm in a blinded fashion wherein the neuropsychologist is not informed at the time of presentation as to whether stimulation is being performed. This allows for a more objective assessment of performance and its relationship (or lack thereof) to stimulation.

An alternate method for establishing pass/fail criteria uses a specified percentage above the baseline error rate as a criterion (Ojemann, 1983b). However, this method assumes that the patient’s error rate remains constant throughout the procedure (which we have found is not always the case) and necessitates numerous trials to establish the percentage. Another method involves clearing a site after establishing that a high proportion of stimulation trials are responded to accurately at a given site (Hamberger, Goodman, Perrine, & Tamny, 2001); however, again this necessitates numerous trials for each function at each electrode pair tested. This approach will prolong the procedure and fatigue the patient.

There may be several points during the mapping where an item is “failed.” At these times, it is important to ascertain that the item is truly failed and not spoiled by other factors. In addition, important clinical information can be gleaned through secondary assessment following the “failed” stimulation trial. If, for example, on a repetition, naming, or comprehension item, a response is not provided during stimulation, after discontinuing stimulation ask the patient to repeat the item or name the object that was presented. It is often enlightening if the patient can then comply with repeating the phrase/question or naming the stimulus item (i.e., naming the pictured object that may still be in visual working memory) that was presented during stimulation, as this helps to validate the language “hit.”

When mapping in the frontal lobe near primary or secondary motor areas, it must be determined that an incorrect or distorted response is not due to motor phenomena rather than language disruption. Speech may sound dysarthric or the patient may report that they “felt something” in their mouth or throat area during stimulation. This can reflect motor contractions as a result of stimulation of the mouth/face area of the motor strip. Have the patient produce sustained nonlinguistic speech sounds that utilize different levels of the speech apparatus (e.g., “kakakakaka” or “lalalalala”) while delivering intermittent stimulation for durations of 1–2 s each. If production of these sounds is disrupted with stimulation, then a motor, rather than a language, area has been identified. When stimulating electrodes that are directly overlying the motor strip or are somewhat more anterior and superior, more obvious motor phenomena are evident (e.g., contralateral facial contraction, forced tongue deviation or retraction). If it is suspected that tongue motor involvement is disrupting the language response, have the patient extend their tongue and look for forced tongue retraction/deviation during stimulation. Occasionally there are very subtle motor effects (due to current spread) which can be “talked through.” That is, the patient may be able to tolerate the motor phenomenon and try to produce the language response despite the stimulation effects on oral motor functioning. It is important to try this when possible as there can be interdigitation between motor and language areas (Lesser, Gordon, & Uematsu, 1994). Attempting language testing at these “soft” motor sites may help to exonerate language cortex in a planned resection zone.

With auditory comprehension, it is important to be mindful of how a patient’s interpretation of their experience can shape their responses. It is not uncommon for patients to report that they did not “hear” the stimulus item that was presented (e.g., during an auditory naming or sentence completion item). In this circumstance, first determine if in fact there is a primary perceptual problem. Often the bandaging around the head can cover the ear and reduce auditory acuity. However, when there is a comprehension deficit due to stimulation, it may be reported by the patient as a “hearing” problem as this may be how the patient interprets their inability to understand the command or sentence.

There can be reasons for item failure that are independent of the language and motor effects of stimulation. Afterdischarges can disrupt cognitive functioning (Lesser et al., 1987). Therefore, when language disruption is noted during stimulation, it is critical that the EEG be monitored to determine whether the impaired response occurred in the context of ADs or whether the EEG immediately post stimulation was “clean.” If there were ADs, then the trial must be repeated until it is determined that the patient can/cannot complete the task in the absence of epileptiform activity. Finally, the cumulative effects of ADs, small seizures, and/or fatigue can progressively decrease the patient’s ability to reliably respond correctly. In this situation, try taking a 10–15 min break. Depending on the health, endurance, and tolerance of the patient, as well as how irritable (hyperexcitable) the cortex within the region of stimulation is, the mapping procedure may need to be discontinued and resumed on another day to achieve maximum reliability and quality of responses. To the greatest extent possible, include this potential need for additional hours (or days) in the pre-resection surgery planning.

Stimulation during continuous speech has been shown to be a sensitive method for identifying language eloquent cortex (Lesser et al., 1987). However, selection of the stimulus material is important. The content of the speech monologue must not contain prosody (such as having the patient sing a song) as this may engage language centers in the contralateral hemisphere. Counting should also not be used for this purpose as knowledge of numbers can be so overlearned that the patient may be able to produce accurate responses even when stimulation of that site would ordinarily disrupt other more complex types of verbal output (Bookheimer, Zeffiro, Blaxton, Gaillard, & Theodore, 2000). Instead, a monologue that is syntactically more complex but is well known to the patient and can be repeatedly produced accurately should be used. For individuals who are born and educated in the United States we often use the Pledge of Allegiance for this purpose. Other types of monologues that can be used include prayers, poems, or lines from a play that are very well known to the patient. These types of passages are selected because they are rote for the patient but not resilient to disruption caused by stimulation, such as can be the case with reciting numbers. If no well-known fixed monologue can be identified for a particular patient, then, as a last resort, the patient can be instructed to speak on a topic of their choice. This latter choice is somewhat risky as it is more difficult for the examiners to identify errors and paraphasias since the words and sentences are not provided in a predetermined and fixed manner. As an alternative, patients can be asked to recite the months of the year or days of the week in a continuous and repeating manner. We tend to use this type of task with lower functioning individuals.

Visual confrontation naming has been used as one of the “gold standards” for identifying language eloquent cortex during mapping since the time the procedure was developed. This type of task has been one of the most sensitive in identifying language eloquent cortex (Ojemann, 1983a). Contrary to the clinical neuropsychology setting where tasks with graduating difficulty are often used (e.g., the Boston Naming Test), in the mapping setting, pictures of items that are easily, rapidly, and consistently named by the individual should be used. As indicated, the clinician should have a large selection of picture stimuli available and then determine at baseline a patient-specific set of approximately ten items that the patient can respond to easily at the time of mapping. We have found the stimuli published by Snodgrass and Vanderwart (1980) to be extremely useful for this purpose.

Although visual naming has been the most common format for assessing word retrieval in the context of brain mapping, Hamberger and others (Hamberger et al., 2001; Lesser et al., 1987) have demonstrated that auditory naming (also known as descriptive or responsive naming) tasks can yield important and unique information. Specifically, it has been established that auditory and visual confrontation naming can be co-localized; however, in many instances these functions can also be distributed across distinct sites in the brain. It could be argued that auditory naming (i.e., retrieving a specific word while you are formulating what you want to say) can be considered more ecologically valid than visual naming, or at least as critical as naming an object in one’s visual field. Thus, it is an important language function to assess during the mapping procedure. Examples of auditory naming items include “what is the thing that unlocks a door” or “what do we tell time with?” At our center, we use the stimuli developed by the Columbia group for this purpose (Hamberger & Tamny, 1999).

In terms of verbal generative fluency, traditional verbal fluency tasks (i.e., continuous productions for words belonging to a specific category, such as those that start with a specific letter) are not widely used in mapping procedures. They can be problematic for the same reason that, for example, continuous extemporaneous speech can be. Specifically, if the patient naturally hesitates in their speech or in their thought process during completion of the task, it could be confused as a “hit,” rather a natural pause. Ojemann and colleagues have used a somewhat related single-word generation task (Ojemann, Ojemann, & Lettich, 2002). Borrowing from the functional imaging literature, they employed a verb generation task. In this task, patients are shown a picture of a concrete noun and asked to generate a verb appropriate to the picture stimulus. Consistent with the robust findings from functional imaging studies, Ojemann et al. (2002) were able show both overlapping and disparate verb generation sites relative to those showing disruption of object naming in the frontal and temporoparietal cortices.