Hearing every point in sequence is a poor substitute for a spoken summary of the overall trend

What it is

A sighted user glances at a line chart and answers "rising or falling, any anomalies?"—a point-by-point listener must hear dozens of points and rebuild the shape mentally before answering the same question. The cost of acquiring the "overall impression" differs by an order of magnitude between the two. A summary layer (a readable overview available at chart entry: "Sales rose overall in 2024, with an anomalous dip in March and a peak in Q4") turns the overall impression from "available after traversal" into "available on arrival"—an auditory-channel equivalent of the graphic's global pattern.

Why it happens

The summary layer's mechanism is cognate with alt text (the same group's text alternatives) but with a different duty: alt text declares the graphic's existence and conclusion to the page's semantic layer, while the summary layer serves users already inside the chart with a structured overview—their content can connect but their granularity differs. The layer's effectiveness matches auditory working memory's capacity: one or two sentences carrying three essentials—direction, extremes, anomalies (rising overall; peak in December; anomalous dip in March)—let users build a global schema in one pass, after which point-by-point traversal becomes "locating within that schema" rather than "building from zero." Without it, point-level information cannot self-assemble into a global judgment (the auditory channel has no afterimage), so users either abandon global understanding or patch it together through repeated traversals. Summary generation can be semi-automated: compute trend direction, extremes, and anomalies from the data, fill a language template for a draft, and have the chart maker review and revise—making the layer sustainable across large-scale chart production rather than a handcrafted luxury.

Where it stops holding

Summary and point traversal complement rather than replace each other: the summary cannot serve precise per-point queries ("what exactly was August?" still needs the point level), and the entry order should be summary first (overview on entry, descend on demand). Generation has a complexity boundary: single-series trend summaries template well, but a multi-series interactive chart's "overall impression" may not fit one sentence (then summarize per series, or state "see series summaries"). The summary's wording must be verified against the data as strictly as a title claim (the same group's announcement quality)—a wrong auto-generated summary is worse than none, because it delivers a wrong global impression from the most authoritative position.

Applying it

• Announce the summary layer first when the chart gains focus: trend direction, extremes, anomalies, in one or two sentences.
• Generate summaries from data-driven templates (direction + extremes + anomaly detection), reviewed by the chart maker before release.
• After the summary, offer an "on-demand depth" path hint ("arrow keys browse points; Tab switches series").
• Verification: have users answer global questions (direction, peak location) after hearing only the summary; their accuracy should substantially exceed users who only traverse point by point—that gap proves the layer's value.

Related

• Same group: U11.02.1 Chart elements need a predictable keyboard traversal order · U11.02.2 Traversal should support switching between series and data-point levels · U11.02.3 Announcements must include identifier, value, and unit · U11.02.5 View changes caused by interaction must be announced
• Nearby: U11.01.1 Graphics need a text summary as equivalent content · U11.02.2 Traversal should support switching between series and data-point levels
• Search terms: trend summary · chart sonification · summary layer