The California Verbal Learning Test (CVLT) is one of the most widely used neuropsychological tests in North America. As an instrument, it represents a relatively new approach to clinical psychology and the cognitive science of memory. It measures episodic verbal learning and memory, and demonstrates sensitivity to a range of clinical conditions. The test does this by attempting to link memory deficits with impaired performance on specific tasks. It assesses encoding, recall and recognition in a single modality of item presentation (auditory-verbal). The CVLT is considered to be a more sensitive measure of episodic memory than other verbal learning tests. It was designed to not only measure how much a subject learned, but also reveal strategies employed and the types of errors made. The CVLT indexes free and cued recall, serial position effects (including primacy and recency), semantic clustering, intrusions, interference and recognition. Delis et al. (1994) released the California Verbal Learning Test for Children (CVLT-C). The California Verbal Learning Test-II (CVLT-II) is an updated version of the original CVLT, which has been standardized and provides normative data.
Measures
CVLT generates a wide variety of measures: Immediate Recall Short Delay Free Recall Short Delay Cued Recall Long Delay Free Recall Long Delay Cued Recall Long Delay Recognition
Administration
The original CVLT was normed on a 'reference sample' of 273 nonclinical subjects.
The experimenter reads a list of 16 nouns aloud, at one-second intervals, in fixed order, over five learning trials (list A). After each trial, the subject is asked to recall as many words as they can in any order (i.e., free recall).
A big feature, compared to other verbal learning tests, is that the words are drawn from four semantic categories (tools, fruits, clothing, spices and herbs), with no consecutive words from the same category. If a subject 'clusters' words from a category together, it is probable that they are using semantic organisation.
An interference list (list B) is presented that shares two categories from List A (e.g., fruit and tools) and has two unshared categories (e.g., fish and kitchen utensils). However, neither list uses common words for a specific category (e.g., apples used rather than bananas). Free and cued recall of list A are tested immediately (short-delay), and again after 20 minutes (long-delay). In cued recall, the experimenter prompts the subjects with the word category.
The CVLT ends with a recognition task, where the experimenter presents the subject with a 44-word list, and the subject must indicate whether it is a target word or a distractor. Some distractors share semantic categories with the target words while others sound alike. The 44-word list is presented like shopping list as it was argued that this is an activity that people face in their everyday activities. Background participant information about age, sex, and ethnicity are recorded for demographic purposes. The words have an average of 2.37 syllables and there are 64% of the items on the recognition list are distractors.
Analysis
A computer administration and scoring system generates scores for every measure, graphs a learning curve, and provides learning parameters, response errors and interference effects. Raw scores are used for all analyses, ultimately determining how many errors are made in each learning task. The Wilcoxon Signed Rank Test assesses practice effects and Spearman's rho (p) correlation coefficient is conducted to assess test-retest reliability. Primary learning, recall and recognition measures are recorded as well as the more detailed process measures such as errors, contrast scores, and ratio.
It is suggested that a general verbal learning component consistently accounts for about 35-40% of the total variance and consists of total free recall over the five trials of list A, semantic clustering free and cued recall (both short- and long-delays), and recognition hits. A second, "response discrimination" component has also been found in most studies. It accounts for about 8-10% of the variance with loadings from free and cued recall intrusions and recognition false positives. The remaining components, learning strategy (semantic and serial clustering), serial position (primacy and recency) and proactive effect (List B recall) are inconsistent and account for little additional variance.