Microaggressions—coded language, exclusionary comments, subtle slights—often hide in professional writing because they're deniable individually. AI trained to recognize these patterns can flag language that human reviewers might normalize or overlook.
Tone is everything in written communication, but it's invisible. An email can contain words that seem normal but convey dismissal, anger, or contempt through how they're phrased. Tone analysis is the process of AI reading a message and identifying the emotional texture behind the words. Think of it like an X-ray for subtext.
For example: "Thanks for the input" can be genuine. "Thanks for the input" (with certain context or phrasing) can be sarcastic. "Thanks for your continued input on this matter" can feel patronizing. A human reader gets this intuitively. AI can be trained to spot it systematically.
Toxic managers often use technically polite language while creating a hostile environment. Your boss can write "I appreciate your perspective, but we handle things differently here" in a way that systematically shuts down every idea you propose. One email seems fine. Fifty of them, read together, show a pattern of dismissal disguised as respect.
Tone analysis helps you document this. Instead of saying "My boss is always dismissive," you can say "In 34 out of 67 emails where I proposed something, my manager used phrases like 'appreciate,' 'but,' and 'different approach'—which linguistically signals disagreement or rejection while appearing polite. The tone is consistent and systematic."
When you feed an AI tool a message, it looks for: passive-aggressive language (indirect criticism), condescending phrases (language that suggests you're less capable), dismissive patterns (shutting down ideas), and contradictions between stated intent and actual language.
For instance, "I value your input, but..." followed by never implementing your input is a tone contradiction. The message claims openness while the behavior shows closure.
AI can spot patterns in language. It can't always detect context that changes meaning. A message meant as a joke might read as hostile. That's why tone analysis is best used alongside your own judgment, not as a replacement for it.
Try this: Take three emails from a workplace relationship you want to analyze. Paste them into ChatGPT and ask: "Analyze the tone of each email. Look for: dismissive language, passive-aggressive phrases, or language that appears polite but contradicts the actual action taken." You'll immediately see patterns you may have felt but couldn't articulate.
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.