Transcranial Magnetic Stimulation (TMS) has been demonstrated to improve language function in subjects with chronic aphasia in a number of small studies, many of which did not include a control group. Although the treatment appears promising, data to date do not permit an adequate assessment of the utility of the technique. The investigators propose to study the effects of TMS combined with Constraint Induced Language Therapy (CILT) in 75 subjects with chronic aphasia. Subjects will be randomized in a 2:1 ratio to TMS with CILT or sham TMS with CILT. One Hz TMS at 90% motor threshold will be delivered to the right inferior frontal gyrus for 20 minutes in 10 sessions over 2 weeks; language therapy will be provided for one hour immediately after the conclusion of each session of TMS. Change from baseline in the Western Aphasia Battery Aphasia Quotient at 6 months after the end of TMS treatment will serve as the primary outcome measure.

TMS is a technique by which a brief electrical current is induced in brain tissue causing a brief suppression of the excitability of the underlying tissue; the technique, which was introduced in the 1980s and has been extensively used around the world, has been shown to transiently improve or disrupt specific cognitive operations. To achieve this end, a coil is positioned against the subject's head. The delivery of a single pulse begins with the discharge of current from a capacitor into a circular or figure-of-eight coil; this electrical current generates a brief magnetic field of up to 2.2 Tesla. As the pulse of electricity has a rise time of 0.2 ms. and a duration of 1 ms., the magnetic field changes in intensity quite rapidly. Because the magnetic field passes freely through the scalp, skull, and meninges, the flux in the magnetic field induces a small electric field in the brain that transiently alters neural activity.

TMS may be delivered in a variety of ways. The investigators propose to use 1 Hz TMS; that is, TMS pulses will be delivered at a frequency of 1/second. This style of TMS is assumed to be inhibitory in that it transiently suppresses the function of the cortex under the coil. Using the figure-of-eight coil to be employed here, TMS is thought to reduce activity in approximately 1 cubic cm. of cortex. Many investigators have employed TMS with a frequency of 1 Hz for periods of 20 minutes and longer; mild behavioral deficits are often present for several minutes in these studies.

The baseline phase will consist of 3 sessions, each lasting 1-2 hours depending on the stamina of the subject. The point of the baseline testing is to characterize the subject's language function. To that end, a number of standard language and neuropsychological tasks will be administered. These include the Western Aphasia Battery, Pyramids and Palm Trees test, Figural Fluency Test, word and non-word repetition tasks, the Nicholas and Brookshire Narratives, CILT stimulus naming, and Northwestern Assessment of Verbs and Sentences. Additionally, during the baseline, subjects will undergo MRI of the brain or, if they have a contraindication to MRI, a CAT scan of the head. No contrast will be used.

In the treatment phase, there will be 10 TMS sessions over 2 consecutive weeks in which 20 minutes (1200 pulses) of 1 Hz TMS at 90% motor threshold will be delivered to the inferior pars triangularis. Each TMS treatment session will be immediately followed by a 60-90 minute session of CILT

There will be two 3-month post-treatment visits and two 6-month post-treatment visits in which the full battery of language and cognitive assessments will be repeated. Subjects who are able to undergo MRI scanning will have anatomic and fMRI scans at the first 6-month post-treatment visit.

The investigators will pair TMS with CILT which has been shown to have positive outcomes in post-stroke aphasia. CILT invokes use-dependent learning in communicative interactions by requiring spoken output and restricting use of alternative forms of communication, such as gestures. The investigators will use a dual card-matching task modeled after Maher et al. As in the original CILT design, the participant interacts verbally with a conversational partner (here, the speech language pathologist), in turn requesting a card of given description and complying with the partner's request. In this way, the treatment targets both production and comprehension. Moreover, as verbal targets increase in linguistic complexity across the protocol ("a ball", "throw a ball"; "Do you have a ball"?), a variety of lexical and phrasal structures are targeted. Studies of CILT have reported gains on multiple language behaviors, supporting its broad engagement of the language network.

Inclusion Criteria:

• Clinical evidence and MRI or CT verification of a single left hemisphere stroke with moderate to severe aphasia.
• Suffered their stroke at least 6 months prior to their testing
• Must be able to understand the nature of the study, and give informed consent

Exclusion Criteria:

• Multiple strokes (excluding small lacunar strokes) as defined by brain imaging
• History of substance abuse
• Previous head trauma with loss of consciousness for more than 5 minutes
• Psychiatric illness (We note that subjects will be assessed with the 15-item Geriatric Depression scale. Because depression is very difficult to evaluate in aphasic subjects, potential subjects will not be excluded on the basis of the depression score)
• Chronic exposure to medications that might be expected to have lasting consequences for the central nervous system (e.g. haloperidol, dopaminergics)
• History of or neuropsychological findings suggestive of dementia

From baseline, adverse event data were collected over a period of time consisting of 6 months.

Adverse event (AEs) information was collected daily during treatment sessions, as well as each follow-up visit.

The individual administering TMS will keep the master file of subject assignments, but all other individuals in contact with the subject or their data will be unaware of group assignment. Participants will not be informed of their assignment to active or sham status.