Behavioral Metrics · 10 min read
Insults, Threats, and Harassment Metrics in Text Messages
How to quantify verbal degradation, profanity used as control, threats, repeated unwanted contact, stalking cues, boundary violations, and escalation windows.
The short answer
These metrics turn repeated verbal aggression into something inspectable rather than something argued about. Instead of asserting that a thread looks bad, the record can show counts and sequences: degrading labels, profanity used to control rather than to vent, direct threats, indirect intimidation, repeated unwanted contact, boundary violations, retaliation language, and the windows in which intensity rose. Two cautions matter. A count measures the supplied record, so it reflects what was exported and says nothing about what was said elsewhere or deleted beforehand. And a category label is a candidate for review, not a finding: whether a message is a threat in the legal sense is a judgement for a person, and ultimately a court, applying a standard the software does not apply. Used properly, the measurement makes a pattern legible and points a reviewer at the messages behind each number, which is where the actual argument happens. This matters for legal professionals, therapy and advocacy contexts, and consumers or survivors because repeated verbal aggression can disappear inside thousands of ordinary messages. A structured metric makes the pattern visible while preserving the source messages that support it.
Metric categories to separate
Insults and degradation — Name-calling, role attacks, humiliation, contempt, profanity directed at identity, parenting, work, body, sobriety, or competence. — Repeated degradation can show a pattern even when no explicit threat appears. Threats — Physical harm, property harm, legal retaliation, reputation damage, financial consequences, immigration or housing leverage. — Severity and target matter for legal, safety, and advocacy review. Harassment — Repeated unwanted messages, message bursts, calls after stop requests, third-party contact, unrelated pressure during logistics. — Frequency and boundary context matter as much as the words. Stalking cues — Location references, showing-up statements, surveillance language, tracking, monitoring, “I saw you” messages. — Source-linked timing helps reviewers separate ordinary knowledge from unwanted monitoring. Boundary violations — Ignoring no-contact requests, pushing visits after refusal, contacting through children or friends, continuing after “stop.” — The boundary message and later conduct need to be reviewed together. Retaliation and evidence control — Delete messages, stop documenting, change your story, I will ruin you, you will regret this. — These messages can affect preservation and credibility questions. Escalation — More frequent messages, more severe language, new threat types, late-night bursts, post-separation surges. — A timeline shows whether the pattern worsened, stabilized, or stopped.
Why context changes severity
A word list alone is not enough. The same profanity can appear in sarcasm, mutual argument, coercive control, or a direct threat. Textimony may surface rule or model candidates, but a reviewer must judge speaker direction, prior boundaries, repetition, timing, target, and surrounding messages. The candidate and context views let a reviewer return to the parsed message instead of trusting a label in isolation. Ambiguous messages should remain unresolved rather than being forced into a conclusion.
Audience-specific value
Legal professionals can use the categories to triage threats, harassment after boundaries, and source messages that may need authentication or context review. Therapists and advocates can use the pattern map as context for client discussions without treating it as diagnosis or automated risk assessment. Consumers and survivors can see whether the issue is isolated insult language, repeated degradation, post-boundary harassment, or escalating threat behavior. Parents can keep co-parenting logistics separate from abusive language so custody issues and harassment issues do not overwrite each other. Reviewers can test whether screenshots fairly represent the larger thread or omit context.
What a careful metric report includes
Counts by category and severity, not one generic negativity score. Date ranges and escalation windows showing when the behavior clustered. Participant mapping and uncertainty notes when sender identity is not fully established. Context windows before and after each selected message. Links to source files, message IDs, attachments, and generated report manifests. Clear source notes for sarcasm, quoted language, missing media, deleted content, or unsupported export fields.
Can insults in text messages matter?
They can, though rarely on their own. An isolated insult during an argument is usually treated as exactly that. Insults carry weight when they are repeated, sustained, targeted at protected characteristics, or paired with threats or controlling behavior — in other words when they form part of a pattern. Context decides it, which is why the surrounding messages matter more than the words themselves.
How does Textimony distinguish insults from harassment?
They are different concepts and Textimony treats them separately. Insult review looks at degrading language in individual messages; harassment review looks at repeated unwanted contact and whether it continued after a boundary was stated, which is a property of the sequence rather than any one message. The software locates candidates for both. Deciding which framing fits the record is a human judgment.
Can a threat be indirect?
Yes, and indirect threats are common precisely because they are harder to point at. A threat may be physical, financial, legal, reputational, housing-related, or tied to custody, and it can be carried entirely by implication — a reference to knowing where someone works, or to what happened to someone else. Context and shared history decide the meaning, so read the source messages before treating a flag as an issue.
Does Textimony decide whether stalking occurred?
No. Stalking is a legal determination with statutory elements that vary by jurisdiction, and it usually depends on conduct outside the messages entirely. Where the category is available, Textimony surfaces candidates containing location references, repeated contact, or language about showing up. Those are places for a person to look, not findings. Legal and safety determinations need qualified review, and immediate danger needs emergency services.
Published by
Textimony. Editorial status: Source-linked informational guide. Updated: 2026-07-12.
Sources
CDC: Stalking and intimate partner violence; The National Domestic Violence Hotline: Types of Abuse; Technology-Facilitated Abuse in Intimate Relationships; Federal Rule of Evidence 901