CoChat

Automated analysis of written language in the three variants of primary progressive aphasia

Open access
Josephy-Hernandez, Sylvia, Rezaii, Neguine, Jones, Amelia, Loyer, Emmaleigh, Hochberg, Daisy, Quimby, Megan, Wong, Bonnie, Dickerson, Bradford C · 2023 · Brain Communications

The DOI matches its Crossref record; 39 of the 40 references it cites were found in a registry or an index.

Abstract

Despite the important role of written language in everyday life, abnormalities in functional written communication have been sparsely investigated in primary progressive aphasia. Prior studies have analysed written language separately in each of the three variants of primary progressive aphasia—but have rarely compared them to each other or to spoken language. Manual analysis of written language can be a time-consuming process. We therefore developed a program that quantifies content units and total units in written or transcribed language samples. We analysed written and spoken descriptions of the Western Aphasia Battery picnic scene, based on a predefined content unit corpus. We calculated the ratio of content units to units as a measure of content density. Our cohort included 115 participants (20 controls for written, 20 controls for spoken, 28 participants with nonfluent variant primary progressive aphasia, 30 for logopenic variant and 17 for semantic variant). Our program identified content units with a validity of 99.7% (95%CI 99.5–99.8). All patients wrote fewer units than controls (P < 0.001). Patients with the logopenic variant (P = 0.013) and the semantic variant (0.004) wrote fewer content units than controls. The content unit-to-unit ratio was higher in the nonfluent and semantic variants than controls (P = 0.019), but no difference in the logopenic variant (P = 0.962). Participants with the logopenic (P < 0.001) and semantic (P = 0.04) variants produced fewer content units in written compared to spoken descriptions. All variants produced fewer units in written samples compared to spoken (P < 0.001). However, due to a relatively smaller decrease in written content units, we observed a larger content unit-to-unit ratio in writing over speech (P < 0.001). Written and spoken content units (r = 0.5, P = 0.009) and total units (r = 0.64, P < 0.001) were significantly correlated in patients with nonfluent variant, but this was not the case for logopenic or semantic. Considering all patients with primary progressive aphasia, fewer content units were produced in those with greater aphasia severity (Progressive Aphasia Severity Scale Sum of Boxes, r = −0.24, P = 0.04) and dementia severity (Clinical Dementia Rating scale Sum of Boxes, r = −0.34, P = 0.004). In conclusion, we observed reduced written content in patients with primary progressive aphasia compared to controls, with a preference for content over non-content units in patients with the nonfluent and semantic variants. We observed a similar ‘telegraphic’ style in both language modalities in patients with the nonfluent variant. Lastly, we show how our program provides a time-efficient tool, which could enable feedback and tracking of writing as an important feature of language and cognition.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/braincomms/fcad202

Cites

39 of 39 references verified0 citation errors · 0 not checkable

Statistics

Figures

Graphical Abstract
Graphical Abstract CC BY · PubMed Central
Figure 1 Comparison of written language between participants with PPA and controls. Graphs illustrate written content units (CUs) (A), total units (U) (B) and CU/U Ratio (C) in controls and the three PPA variants. Error bars indicate 1 SD. Statistically significant differences indicated by lines correspond to one-way ANOVA with Tukey’s post hoc tests.
Figure 1 Comparison of written language between participants with PPA and controls. Graphs illustrate written content units (CUs) (A), total units (U) (B) and CU/U Ratio (C) in controls and the three PPA variants. Error bars indicate 1 SD. Statistically significant differences indicated by lines correspond to one-way ANOVA with Tukey’s post hoc tests.CC BY · PubMed Central
Figure 2 Comparison of written and spoken languages within participants with PPA. Each graph shows the mean and 1 SD for the group of participants on the left (written) and right (spoken) of each panel, with individual datapoints from the same participant connected by a line in the middle. Data from each of the PPA variants are displayed separately. The rows of graphs show content units (CUs) (A), total units (B) and CU/U Ratio (C). The P-values at the top of each panel correspond to Tukey’s HSD post hoc contrasts, in a two-way one-within/one-between ANOVA comparing the written to spoken modality between each PPA subtype.
Figure 2 Comparison of written and spoken languages within participants with PPA. Each graph shows the mean and 1 SD for the group of participants on the left (written) and right (spoken) of each panel, with individual datapoints from the same participant connected by a line in the middle. Data from each of the PPA variants are displayed separately. The rows of graphs show content units (CUs) (A), total units (B) and CU/U Ratio (C). The P-values at the top of each panel correspond to Tukey’s HSD post hoc contrasts, in a two-way one-within/one-between ANOVA comparing the written to spoken modality between each PPA subtype.CC BY · PubMed Central
Figure 3 Scatterplots comparing spoken and written CUs, total units and CU/U Ratio. In all graphs, written language is on the y-axis, and spoken language is on the x-axis. The left column shows CU, the middle column shows total units and the right column shows CU/U Ratio. The rows separate the three PPA variants, with the bottom row showing the three variants combined. Statistics included refer to Bivariate Pearson correlation analysis.
Figure 3 Scatterplots comparing spoken and written CUs, total units and CU/U Ratio. In all graphs, written language is on the y-axis, and spoken language is on the x-axis. The left column shows CU, the middle column shows total units and the right column shows CU/U Ratio. The rows separate the three PPA variants, with the bottom row showing the three variants combined. Statistics included refer to Bivariate Pearson correlation analysis.CC BY · PubMed Central
Figure 4 Correlation analysis of written and spoken CUs and total units from all participants with PPA against their respective clinical rating scores. The left column corresponds to content units (CUs) and the right to total units. The top four boxes correspond to the written modality and the bottom four to spoken. Each modality is further subdivided into correlation with the Clinical Dementia Rating scale Sum of Boxes (CDR SoB) or the Progressive Aphasia Severity Scale Sum of Boxes (PASS SoB) as shown on the right. Statistics shown correspond to Bivariate Pearson correlation analysis.
Figure 4 Correlation analysis of written and spoken CUs and total units from all participants with PPA against their respective clinical rating scores. The left column corresponds to content units (CUs) and the right to total units. The top four boxes correspond to the written modality and the bottom four to spoken. Each modality is further subdivided into correlation with the Clinical Dementia Rating scale Sum of Boxes (CDR SoB) or the Progressive Aphasia Severity Scale Sum of Boxes (PASS SoB) as shown on the right. Statistics shown correspond to Bivariate Pearson correlation analysis.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Cited by

See all 26

Metadata

Authors
Josephy-Hernandez, S., Rezaii, N., Jones, A., Loyer, E., Hochberg, D., Quimby, M., Wong, B., & Dickerson, B. C.
Affiliations
  • Frontotemporal Disorders Unit, Department of Neurology, Massachusetts General Hospital & Harvard Medical School , Boston, MA 02114 , USA
  • Harvard University (US)
  • Spaulding Rehabilitation Hospital (US)
Corresponding author
Dickerson, Bradford C
Venue
Brain Communications
Published
2023
Volume
5
Issue
4
Pages
fcad202-fcad202
ISSN
2632-1297 (ISSN-L)
OpenAlex
W4384930306
Cited by
25
References
40
Citations by year
2026: 6 · 2025: 10 · 2024: 6 · 2023: 2
FWCI
4.87
Type
journal article
Language
en
Open access
Locations
  • Brain Communications (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
Funders
  • Sidney R. Baer Jr. Foundation
  • National Institutes of Health (R01 DC014296, R21 AG073744, R21 DC019567)
  • Tommy Rickles Endowed Chair
  • Primary Progressive Aphasia Research
Topics
  • Health Sciences › Medicine › Psychiatry and Mental health › Dementia and Cognitive Impairment Research
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neurobiology of Language and Bilingualism
  • Health Sciences › Health Professions › Occupational Therapy › Assistive Technology in Communication and Mobility
Keywords
primary progressive aphasia, written language, logopenic variant, semantic variant, Western Aphasia Battery, aphasia severity, dementia severity, written vs. spoken language, computational text analysis, dementia severity assessment, Clinical Dementia Rating

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.