Boston Naming Test

1. Abstract

The Boston Naming Test (BNT; including its Dutch adaptations, the Boston Benoemingstest [BBT] and the revised Nederlandse Benoem Test [NBT]) represents the gold-standard neuropsychological instrument for the assessment of visual confrontation naming and word-retrieval integrity. Originally formulated by Harold Goodglass, Edith Kaplan, and Sandra Weintraub in 1983 as a specialized subcomponent of the Boston Diagnostic Aphasia Examination (BDAE), the BNT is composed of 60 line drawings arranged in order of ascending psycholinguistic difficulty. The instrument assesses the cognitive mechanisms underlying lexical-semantic retrieval, phonological assembly, and communicative output. Administered individually, respondents are instructed to name each visually presented referent within a 20-second latency window, with standardized semantic (stimulus) and phonemic cues supplied hierarchically following naming failures or visual misperceptions.

Psychometrically, the BNT demonstrates exceptional internal consistency (Cronbach’s alpha and split-half reliability coefficients routinely exceeding .85 to .97 across diverse neurological cohorts) and high test-retest reliability ($r > .89$). Structural and construct validity investigations confirm that the test operates primarily across a unidimensional confrontation naming continuum, heavily mediated by target word frequency, semantic familiarity, visual complexity, and age of acquisition. The instrument is sensitive to dysnomia across a spectrum of etiologies, including cerebrovascular accidents (stroke), Alzheimer’s disease, frontotemporal lobar degeneration, traumatic brain injury, epilepsy, and Parkinson’s disease. Furthermore, the Dutch standardized editions (van Loon-Vervoorn & Stumpel, 1996; and the updated NBT, 2018) provide robust clinical norms adjusting for age and educational attainment, solidifying the instrument’s indispensable role in both clinical diagnostic formulations and empirical cognitive neuroscience.

2. Keywords

Boston Naming Test, Boston Benoemingstest, confrontation naming, anomia, word retrieval, aphasia, neuropsychological assessment, lexical access, semantic memory, visual agnosia, psycholinguistics, cognitive impairment

3. Authors

The original Boston Naming Test was developed by prominent clinical neuropsychologists and neurolinguists at the Boston Veterans Administration Medical Center and Boston University School of Medicine:

• Harold Goodglass, Ph.D. (1920–2002): Co-founder of the Harold Goodglass Aphasia Research Center, Department of Neurology, Boston University School of Medicine; pioneer in the cognitive classification of aphasic syndromes.
• Edith Kaplan, Ph.D., ABPP (1924–2009): Professor of Neurology and Psychiatry, Boston University School of Medicine; originator of the “Boston Process Approach” emphasizing qualitative error analysis alongside quantitative psychometrics.
• Sandra Weintraub, Ph.D., ABPP: Professor of Psychiatry and Behavioral Sciences and Neurology, Mesulam Center for Cognitive Neurology and Alzheimer’s Disease, Northwestern University Feinberg School of Medicine.

The standard Dutch adaptation (Boston Benoemingstest) was executed by:

• W. A. van Loon-Vervoorn, Ph.D.: Department of Psychonomics and Experimental Psychology, Utrecht University, Netherlands.
• H. J. Stumpel, M.Sc.: Clinical Neuropsychology Section, Rehabilitation Center De Trappenberg and Utrecht University.

The modernized 2018 Dutch iteration (Nederlandse Benoem Test, NBT) was authored by specialized clinical linguists and neuropsychologists focused on acquired brain injury (Niet Aangeboren Hersenletsel [NAH]).

4. Purpose

The primary purpose of the Boston Naming Test is the quantitative and qualitative evaluation of visual confrontation naming performance, serving as an exquisitely sensitive index for detecting anomia (dysnomia)—one of the most pervasive, debilitating, and earliest manifestations of linguistic disruption resulting from focal or diffuse cerebral pathology. Initially devised to complement the Boston Diagnostic Aphasia Examination (BDAE), the BNT addresses the critical clinical necessity to identify mild, subclinical, or emergent word-finding difficulties that often escape detection on gross conversational speech evaluations or omnibus cognitive screening batteries.

From a clinical diagnostic standpoint, confrontation naming requires the precise synchronization of multiple neurocognitive processing tiers: intact visual apperception, structural representation matching, semantic memory access, lexical lemma selection, phonological node retrieval, and coordinated motor articulation. Consequently, the instrument acts as a diagnostic bellwether across an expansive spectrum of neurological conditions. In patients recovering from acute stroke or intracerebral hemorrhage, the BNT delineates the boundary between classic perisylvian aphasias (e.g., Broca’s, Wernicke’s, and conduction aphasias) and transcortical or primary anomic variants. In geriatric and neurodegenerative settings, the BNT demonstrates high diagnostic specificity in distinguishing amnestic mild cognitive impairment (MCI) and early-stage Alzheimer’s disease—where degraded semantic architecture produces disproportionate confrontation naming deficits—from normal cognitive aging or frontostriatal disorders such as Parkinson’s disease and progressive supranuclear palsy.

Beyond dichotomous diagnostic classification, the BNT affords profound utility in longitudinal patient monitoring and surgical planning. It is extensively utilized in pre- and post-operative evaluations for resective neurosurgery in drug-resistant temporal lobe epilepsy and eloquent cortex tumor resection, frequently serving as the functional foundation during intraoperative awake cortical stimulation mapping. In academic and cognitive research settings, the Dutch editions (BBT and NBT) have been employed to chart lexical acquisition and vocabulary enrichment trajectories across childhood development, unravel cognitive reserve dynamics in aging, and dissect lexical access architectures within monolingual and multilingual cohorts.

5. Psychological Construct

The psychological construct assessed by the Boston Naming Test is visual confrontation naming, defined psycholinguistically as the rapid, accurate retrieval and phonological emission of an unambiguous spoken label corresponding to a visually presented two-dimensional line drawing. Although confrontation naming is frequently conceptualized as a singular communicative skill, contemporary cognitive neuropsychology models it as an intricate, multi-stage processing pipeline composed of discrete cognitive-perceptual subprocesses:

Visual Structural Encoding

The initial phase involves the bilateral occipitotemporal extraction of perceptual primitives (edges, shapes, volumetric properties) and their synthesis into an integrated structural representation. Failure at this stage produces visual agnosia rather than true anomia; an individual cannot recognize the depicted stimulus, mistaking a tripod for an animal or an abacus for a fence. The BNT systematically controls for this processing level by providing standardized semantic stimulus cues (e.g., for item 50 Pyramid: “It is found in Egypt”) whenever a patient misconstrues the pictorial representation.

Semantic Memory Activation

Upon structural recognition, the percept triggers the activation of a distributed network of conceptual knowledge situated predominantly within polymodal cortical zones, especially the anterior and inferior temporal lobes. This conceptual node contains the semantic attributes, functional utilities, ecological contexts, and associative properties of the object (e.g., recognizing that an Accordion is a musical instrument with bellows and keys). Patients with degraded semantic memory stores, as seen in semantic variant primary progressive aphasia (svPPA), fail at this juncture, often demonstrating preserved perceptual processing (e.g., accurately describing visual attributes) but displaying profound inability to access conceptual associations.

Lexical-Semantic Access (Lemma Selection)

Once semantic activation surpasses an activation threshold, it engages the lemma selection stage, localized primarily within the left middle temporal and angular gyri. The lemma corresponds to the abstract, non-phonological grammatical entry of the target word. Pathologies interrupting this node produce the prototypical “tip-of-the-tongue” phenomenon or circumlocutory output, wherein the patient retains full semantic knowledge (e.g., stating for Stethoscope: “Doctors wear it around their neck to listen to the heart and lungs”) yet fails to isolate the exact lexical entry.

Phonological Form Retrieval and Articulatory Assembly

Following lemma selection, the appropriate phonological code (the metrical frame and individual phonemes) is retrieved from left superior temporal and inferior parietal architectures (Wernicke’s area and supramarginal gyrus) and subsequently transferred to the left inferior frontal gyrus (Broca’s area) and ventral premotor cortex for motor-articulatory programming. Failures at this phase manifest as phonemic paraphasias (e.g., producing “pelican” as “petican” or “cannoe” as “tanoe”). The BNT’s administration protocol directly isolates this construct level through the provision of phonemic cues (supplying the initial sound or syllable), which typically resolves retrieval deficits in conduction aphasia and frontal executive dysnomias while providing minimal assistance in profound semantic degradation.

6. Theoretical Framework

The theoretical framework grounding the Boston Naming Test is anchored in classical cognitive neuropsychology, specifically informed by the connectionist, modular models of language processing developed by Harold Goodglass and Norman Geschwind, along with subsequent psycholinguistic models formalized by Levelt, Roelofs, and Meyer (1999) and Dell’s Interactive Activation Model (1986).

The Wernicke-Lichtheim-Geschwind Disconnection Paradigm

The original theoretical foundation of the BNT rests upon classical behavioral neurology. According to this framework, confrontation naming is an emergent property of distributed cortical centers interconnected via white matter fasciculi—most notably the arcuate fasciculus and the inferior longitudinal fasciculus. Visual inputs processed within primary and association visual cortices (Brodmann areas 17, 18, 19) must be conveyed to polymodal association areas (angular gyrus, area 39) to elicit semantic activation, relayed to Wernicke’s area (area 22) for acoustic-phonetic representation, and forwarded anteriorly to Broca’s area (areas 44 and 45) for phonetic encoding and articulatory implementation. Under this framework, distinct naming errors pinpoint anatomical and functional disconnections along this neuroanatomical axis.

Levelt’s Spreading Activation and Discrete Two-Stage Model

In modern psycholinguistics, Willem Levelt’s WEAVER++ computational architecture provides the theoretical underpinning for interpreting BNT performance. Levelt posits two discrete, non-overlapping serial stages: Conceptual-to-Lemma Selection and Lemma-to-Phonological Form Encoding. Confrontation naming begins with the activation of a conceptual representation in semantic space. Activation spreads across neighboring nodes; the target concept accumulates the highest activation, triggering the selection of a single grammatical lemma. Only after lemma selection does phonological encoding commence. The 20-second latency window implemented in the BNT accommodates the temporal dynamics of this spreading activation cascade, allowing clinicians to distinguish between physiological retrieval delays, complete semantic blockades, and motor execution latencies.

Item Gradient and the Psycholinguistic Difficulty Continuum

A foundational theoretical tenet of the BNT is that confrontation naming operates along an escalating continuum of difficulty determined by psycholinguistic parameters. The 60 items are deliberately sequenced according to descending objective lexical frequency (e.g., high-frequency targets like Bed and Tree appear first, transitioning toward exceptionally low-frequency items like Protractor, Abacus, and Sextant). Concurrently, this sequence inversely correlates with Age of Acquisition (AoA), imageability, and visual prototypicality. By engineering the test as an ordinal psycholinguistic gradient, Goodglass and Kaplan ensured that mild word-finding deficits could be unmasked at the tail end of the test without subjecting severely impaired patients to immediate failure and catastrophic reaction.

7. Validity

The Boston Naming Test has undergone extensive psychometric validation across five decades, establishing an exhaustive empirical profile across construct, criterion-related, convergent, and discriminant validity paradigms.

Construct and Convergent Validity

Construct validity is evidenced by exceptionally robust correlations between BNT total scores and established clinical indices of linguistic and cognitive functioning. Studies consistently demonstrate high convergent correlations ($r = .75$ to $.88$) between the BNT and the Word Discrimination and Oral Expression subscales of the Boston Diagnostic Aphasia Examination (BDAE), the Western Aphasia Battery (WAB-R) Naming and Aphasia Quotients, and the Multilingual Aphasia Examination (MAE) Visual Naming subtest. When evaluated against nonverbal semantic batteries, such as the Pyramids and Palm Trees Test (PPTT), the BNT exhibits moderate-to-strong correlations ($r = .60$ to $.74$), reflecting shared semantic memory variance while preserving clear conceptual divergence from purely non-linguistic visual associative tasks.

Discriminant Validity

Discriminant validity has been rigorously demonstrated across healthy aging cohorts and diverse neurological populations. BNT scores correlate modestly or weakly ($r < .30$) with measures of visuospatial construction (e.g., Rey-Osterrieth Complex Figure copy, Judgment of Line Orientation) and primary sensory-motor thresholds, confirming that performance is not an artifact of gross visuoperceptual or mechanical failure. Crucially, the BNT successfully dissociates normal cognitive decline from neurodegenerative dementias. In ROC (Receiver Operating Characteristic) curve analyses, the BNT yields Area Under the Curve (AUC) values ranging from .89 to .96 in discriminating healthy older controls from patients with mild probable Alzheimer’s disease, with sensitivity and specificity commonly exceeding 85% at optimal normative cut-offs.

Predictive and Ecological Validity

In clinical outcome studies, BNT performance serves as a powerful predictive marker. In acute post-stroke rehabilitation, baseline BNT scores reliably predict long-term functional communicative independence and return-to-work capacity measured at 6- and 12-month follow-ups. In neurosurgical contexts, baseline BNT naming efficiency and error distributions accurately predict post-resective speech preservation in left dominant anterior temporal lobectomies.

8. Reliability

The psychometric reliability of the Boston Naming Test has been replicated across multiple languages, normative samples, and neurological clinical populations, establishing high internal consistency, test-retest stability, and inter-examiner concordance.

Internal Consistency

Across the complete 60-item configuration, estimates of internal consistency are exceptionally high. Initial standardizations and subsequent large-scale normative studies (e.g., Tombaugh & Hubley, 1997; Mack et al., 1992) reported Cronbach’s alpha coefficients ranging between .88 and .96 in healthy adult cohorts, and surpassing .97 in heterogeneous aphasic cohorts. Split-half reliability, calculated utilizing odd-even item dichotomies and adjusted via the Spearman-Brown prophecy formula, yields coefficients consistently situated between .86 and .94. In the Dutch normative investigations of the Boston Benoemingstest (van Loon-Vervoorn & Stumpel, 1996), internal consistency measures aligned tightly with the original English standards, yielding alpha coefficients between .87 and .91 across neurologically intact adults and stroke patients.

Test-Retest Stability

The temporal stability of the BNT has been evaluated across varying inter-assessment intervals. In clinically stable neurological cohorts and healthy older adults re-evaluated across intervals of two to four weeks, test-retest reliability coefficients routinely fall between $r = .89$ and $r = .95$. Over extended intervals (e.g., one to two years in longitudinal cognitive aging studies), stability remains robust ($r = .78$ to $.84$), with modest practice effects typically confined to healthy individuals who exhibit slight naming latency reductions rather than substantial ceiling score shifts.

Inter-Rater and Scorer Reliability

Because the BNT requires the clinical examiner to make online decisions regarding the acceptability of approximate word forms, regional dialectal variations, visual misperceptions, and phonemic distortions, inter-rater reliability is vital. Studies utilizing blinded dual-scoring of audio- and video-recorded administrations report inter-rater agreement exceeding 96%, with intraclass correlation coefficients (ICC) ranging from .94 to .99. Standardized administration protocols—which provide exhaustive lists of acceptable vernacular alternates, visual cue requirements, and explicit phonemic prompts—largely mitigate scoring variability across clinicians.

9. Factor Analysis

Extensive exploratory (EFA) and confirmatory factor analyses (CFA) have examined the latent dimensional architecture of the 60-item Boston Naming Test. Despite the psycholinguistic diversity of the items, empirical factor analytic investigations overwhelmingly support an essentially unidimensional structural model of visual confrontation naming, alongside secondary orthogonal or hierarchical factors dictated by item-specific psycholinguistic properties.

Unidimensionality and Item Response Theory (IRT)

Early EFA investigations employing principal component extraction regularly isolate a dominant first factor accounting for approximately 40% to 65% of the total item variance in clinical samples, with an eigenvalue far exceeding that of subsequent factors (often possessing an eigenvalue ratio greater than 5:1). Item Response Theory (IRT) modeling, specifically the application of two-parameter logistic (2PL) and Rasch models, confirms that the items function along a singular latent trait continuum ($ heta$) representing confrontation naming ability. Items occupying lower indices (e.g., Bed, Tree, Peach) exhibit negative location/difficulty parameters ($eta < -2.0$), identifying them as high-probability endorsements indicative of minimal impairment, whereas terminal items (e.g., Abacus, Protractor, Sextant) exhibit difficulty parameters surpassing $eta > +2.0$.

Multifactorial Exploratory Models

When multi-factor solutions are extracted and rotated (e.g., via Promax or Varimax methods), the emergent latent factors consistently reflect psycholinguistic categorical clusters and stimulus familiarity gradients rather than divergent cognitive processes:

• Factor 1: High-Frequency / Living Referents: Dominated by items such as Tree, Flower, Peach, Camel, Pelican, and Octopus. This factor is characterized by high semantic familiarity, biological categorization, and early age of acquisition.
• Factor 2: Low-Frequency / Artifacts and Mechanical Tools: Characterized by items such as Sphinx, Yoke, Trellis, Trestle/Tripod, Protractor, and Sextant. Item loadings on this dimension reflect formal educational attainment, specialized cultural knowledge, and mechanical utility.
• Factor 3: Culturally Dependent / Low Prototypicality Items: Items such as Scroll, Palette, Abacus, and Dominoes load onto this subsidiary factor, which often captures variance related to visual complexity and cohort-specific historical familiarity.

Structural Equation Modeling (SEM) and Model Fit

Confirmatory factor analyses testing a bifactor structure (comprising a general confrontation naming factor alongside specific biological versus artifact group factors) have demonstrated acceptable-to-superior goodness-of-fit indices across large adult samples: Comparative Fit Index ($ ext{CFI}) ge .93$, Tucker-Lewis Index ($ ext{TLI}) ge .92$, Root Mean Square Error of Approximation ($ ext{RMSEA}) le .048$ ($90%\text{ CI } [.042, .054]$), and Standardized Root Mean Square Residual ($ ext{SRMR}) le .052$. These indices confirm that while minor item-level sub-clusters exist, the total spontaneous score functions as an empirically defensible, unified metric of confrontation naming capacity.

10. Instrument / Measurement Tool

The structural and procedural parameters of the Boston Naming Test are outlined below:

• Test Type: Neuropsychological visual confrontation naming test; individual administration.
• Format: Stimulus booklet containing 60 standardized black-and-white line drawings presented sequentially. In the 2018 Dutch adaptation (Nederlandse Benoem Test, NBT), modern, culturally calibrated color/monochrome photographic and unambiguous vector stimuli are utilized.
• Number of Items: 60 items arranged in an escalating order of psycholinguistic difficulty.
• Administration Time: Approximately 10 to 20 minutes (extending up to 30 minutes in patients with severe dysphasia or motor-articulatory impairments).
• Target Population: Children, adolescents, working-age adults, and geriatric populations; particularly individuals presenting with cerebrovascular accidents (CVA/stroke), traumatic brain injury (TBI), non-congenital brain injury (Niet Aangeboren Hersenletsel [NAH]), neurodegenerative dementias (Alzheimer’s disease, FTLD, Lewy body disease), epilepsy, and brain neoplasm.
• Response Scale: Dichotomous naming accuracy (1 = Correct spontaneous naming within 20 seconds; 0 = Incorrect/no response, followed by standardized stimulus/phonemic cues).
• Standardized Cuing Hierarchy:
  • Stimulus Cue (Semantic/Visual): Provided when a failure appears attributable to visual misperception (e.g., misinterpreting Dart as a pen) or when the participant demonstrates complete lack of visual recognition. If the subject names the item correctly following a stimulus cue, the item is credited in the adjusted cue score, but not in the spontaneous naming tally.
  • Phonemic Cue: Provided immediately upon naming failure, refusal, or persistent erroneous response after the stimulus cue. The examiner provides the initial phonetic sound or syllable of the target word (e.g., “kuh” for Canoe). Phonemic cue successes are systematically tallied to gauge phonological retrieval access, but they receive a score of 0 on spontaneous naming accuracy.
• Basal and Ceiling Rules: In the standard administration protocol for non-aphasic clinical populations, testing may start at item 30 (Escalator). If the patient names items 30 through 37 without error, items 1 through 29 are credited as correct (basal rule). If an error occurs between items 30 and 37, testing proceeds backwards to item 1 until 8 consecutive unassisted correct responses are established. A ceiling rule terminates administration following 6 to 8 consecutive failed items without successful spontaneous naming.
• Scoring Rules:
  • Spontaneous Correct Score: Sum of all items named correctly without cues within 20 seconds (Maximum = 60).
  • Total Correct Score: Spontaneous correct responses plus correct responses following a stimulus cue (Maximum = 60).
  • Phonemic Cue Advantage: Total number of items correctly emitted following phonemic prompting (indexed separately as a measure of phonological facilitation).
  • Error Analysis: Qualitative tracking of semantic paraphasias (within-category or superordinate substitutions), phonemic paraphasias, neologisms, perseverations, and circumlocutions.

11. Permissions & Fee and Test Year

The development and commercial distribution history of the instrument spans multiple decades:

• Publication History: The experimental version emerged in 1978; the definitive first commercial edition was released in 1983 by Harold Goodglass, Edith Kaplan, and Sandra Weintraub. The second edition (BDAE-2 / BNT-2) was published in 2001. The standardized Dutch version (Boston Benoemingstest, BBT) was published in 1996 by W. A. van Loon-Vervoorn and H. J. Stumpel. The modernized Dutch confrontation test (Nederlandse Benoem Test, NBT) was introduced in 2018.
• Copyright & Commercial Distribution: The English Boston Naming Test is copyrighted and commercially distributed by PRO-ED Inc. and Pearson Clinical Assessment. The Dutch editions (BBT and NBT) are distributed through clinical test publishers in the Netherlands and Belgium (such as Pearson Assessment Benelux and Hogrefe Uitgevers).
• Permissions & Fees: The BNT, BBT, and NBT are proprietary clinical instruments. Standard stimulus books, clinical record forms, and administration manuals must be purchased from authorized publishers. Academic researchers seeking to reproduce items or digitize stimuli for experimental protocols must secure formal licensing agreements and written copyright permission from the respective publishing holders.

12. References

Dell, G. S. (1986). A spreading-activation theory of retrieval in sentence production. Psychological Review, 93(3), 283–321. https://doi.org/10.1037/0033-295X.93.3.283

Goodglass, H., & Kaplan, E. (1983). The assessment of aphasia and related disorders (2nd ed.). Lea & Febiger.

Goodglass, H., Kaplan, E., & Weintraub, S. (1983). Boston Naming Test. Lea & Febiger.

Goodglass, H., Kaplan, E., & Barresi, B. (2001). The Boston Diagnostic Aphasia Examination (3rd ed.). Lippincott Williams & Wilkins.

Kaplan, E., Goodglass, H., & Weintraub, S. (2001). Boston Naming Test (2nd ed.). PRO-ED.

Levelt, W. J., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioral and Brain Sciences, 22(1), 1–38. https://doi.org/10.1017/s0140525x99001776

Mack, W. J., Freed, D. M., Williams, B. W., & Henderson, V. W. (1992). Boston Naming Test: Shortened versions for use in Alzheimer’s disease. Journal of Gerontology, 47(3), P154–P158. https://doi.org/10.1093/geronj/47.3.p154

Tombaugh, T. N., & Hubley, A. M. (1997). The 60-item Boston Naming Test: Normative data stratified by education and age. Journal of Clinical and Experimental Neuropsychology, 19(6), 922–932. https://doi.org/10.1080/01688639708403773

van Loon-Vervoorn, W. A., & Stumpel, H. J. (1996). Boston Benoemingstest. Lisse: Swets & Zeitlinger.

van Loon-Vervoorn, W. A., van der Velden, E., & Stumpel, H. J. (2018). Nederlandse Benoem Test (NBT). Houten: Bohn Stafleu van Loghum.

13. Items of the Scale