Clang Associations: Examples, Definition and Coding

Clang associations are what happens when the sound of a word, not its meaning, chooses the next word. The speaker says “sense,” and “cents” arrives, and then “dollars,” and the reply has moved somewhere nobody intended. Andreasen’s Scale for the Assessment of Thought, Language, and Communication (TLC) defines clanging as “a pattern of speech in which sounds rather than meaningful relationships appear to govern word choice, so that the intelligibility of the speech is impaired and redundant words are introduced,” and adds that it “may also include punning associations, so that a word similar in sound brings in a new thought” (Andreasen, 1986).

Two things make clang associations worth a guide of their own: they are among the more distinctive signs in the thought-disorder vocabulary, and they are one of the lower-agreement TLC items (weighted κ = 0.58). Clanging speech is the label most textbooks use for the same thing.

What are clang associations?

Clang associations are a pattern of speech in which the sound of words governs their selection, so that rhymes, puns and sound-alikes pull the speaker from one thought to the next and intelligibility suffers (Andreasen, 1986). The TLC uses the single word clanging for the item. The definition has three parts that all have to be present: sound is doing the choosing, the speech becomes harder to understand, and redundant words come in that meaning alone would not have produced.

The TLC’s published example shows all three. “I’m not trying to make noise. I’m trying to make sense. If you can make sense out of nonsense, well, have fun.” and “I’m trying to make sense out of sense. I’m not making sense [cents] anymore. I have to make dollars” (Andreasen, 1986). Noise rhymes toward sense, sense toward nonsense, and then the pun on cents produces dollars, a word the meaning of the sentence never called for. The thought has been steered by sound.

Clanging sits in the TLC’s first group of eleven items, the ones Andreasen grouped as “more suggestive of severe psychopathology,” alongside derailment, incoherence and pressure of speech, and it has a twin in the SAPS positive thought disorder subscale under the same name (Andreasen, 1986). The full vocabulary is covered in our guide to the Thought, Language and Communication scale.

What do clang association examples sound like?

Clang associations sound like three mechanisms at work: rhyme choosing the next word, a pun carrying the speaker to a new thought, and a sound-alike substituted for the word meaning called for (Andreasen, 1986). The replies below are synthetic, written to show each.

Interviewer: How has work been?

Speaker: Work is a jerk, a perk, I’d rather lurk, the clerk said so, Clark Kent, can’t pay the rent, the rent is bent.

Interviewer: Did you take the medication this morning?

Speaker: Morning, mourning, warning, the pills are hills, Bill paid the bills, I’m ill at will, still.

Interviewer: What would you like to do this week?

Speaker: Week, weak, I’m at the peak, the creek, I speak Greek, no, I don’t, the boat has a leak and the week is bleak.

In the first reply, “work” chooses “jerk,” “perk” and “lurk” by rhyme, “clerk” opens a pun that drags in Clark Kent, and “Kent” chooses “rent.” Nothing about the job was said. In the second, “morning” and “mourning” are a sound-alike substitution, and “pills” sets off a run of rhymes that ends in a word, “still,” with no job in the sentence at all. That is the “redundant words” clause of the definition. The third includes a self-correction (“no, I don’t”), which does not cancel the rating: the sound still chose the words.

A severity rating follows the TLC anchors, which count occurrences across an interview of roughly 50 minutes: mild when clanging occurs once, moderate at two to four times, severe at five to ten, and extreme when it occurs more than ten times or so often that the interview is incomprehensible (Andreasen, 1986). One clang in a long interview is a mild rating, not a feature of the person’s speech.

How do you tell clanging from puns, neologisms and derailment?

Clanging is separated from its neighbors by what is steering the word choice and by what the result does to meaning. A pun chosen for effect keeps the thread; a neologism is a new word rather than a sound-chosen old one; derailment slides between ideas without any sound link. The table lines them up.

Pattern	What chooses the next word	Is meaning preserved?	TLC rating
Clanging (clang associations)	The sound of the previous word: rhyme, pun, sound-alike	No; intelligibility impaired, redundant words appear	Clanging
Deliberate pun, joke, lyric, rhyme for effect	The speaker, on purpose	Yes; the thread continues	Not rated
Neologism	A new, invented word	Partly; the word has no shared meaning	Neologisms (separate item)
Word approximation	An old word used in a new, unconventional way	Partly; meaning can often be guessed	Word approximations (separate item)
Derailment (flight of ideas)	An oblique association between ideas, not sounds	Each sentence yes; the chain no	Derailment
Perseveration	Persistent repetition of the same word or idea	Yes, but stuck	Perseveration (separate item)

The hardest boundary is with ordinary wit. Puns and rhymes are everywhere in healthy speech, especially under pressure to be entertaining, and the TLC’s definition protects against over-rating them with two conditions: intelligibility must be impaired and redundant words must be introduced. A pun that lands and leaves the conversation where it was is not clanging. A pun that abducts the sentence is.

The boundary with derailment runs the other way. Clanging often rides on pressured speech in mania, where it is one of several signs that travel together; the related pair is covered in pressured speech vs flight of ideas. When ideas slip with no sound link, rate derailment. When the slip happened on a rhyme or a pun, rate clanging, and rate derailment as well if the ideas also moved; the TLC scores each item separately, so one stretch of speech can carry two ratings (our reading, not an explicit TLC rule). The sentence-level breakdown of word salad is a third thing again, covered in word salad vs derailment.

Are clang associations a sign of mania or schizophrenia?

Clanging leans toward mania, and it is not diagnostic on its own. In Andreasen’s comparison of patients with mania, depression and schizophrenia, clanging was one of four signs more strongly associated with mania, with pressure of speech, distractible speech and circumstantiality; derailment, tangentiality, incoherence, illogicality and loss of goal did not differ between mania and schizophrenia (Andreasen, 1979b; we checked this against secondary summaries, as the full text was not available to us). That is why clang associations are classically listed among the speech signs of mania, and why a rater who hears them should look for the company they keep: pressure, distractibility and a rapid slide between ideas.

The TLC paper is also explicit that language behaviors “such as associative loosening, clanging, blocking, over-concrete or repetitive speech, and poverty of speech also occur in other psychiatric disorders such as mania or depression, and they also occur in the speech of people who do not meet the criteria for any psychiatric diagnosis, particularly when they are fatigued or stressed” (Andreasen, 1986). A rater describes the sign; the clinician with the whole picture decides what it means. Where mania itself is being rated, the Young Mania Rating Scale rates speech (rate and amount) and language-thought disorder as separate items, and the TLC sign feeds that judgment rather than replacing it.

How do you code clang associations from a transcript?

Code each clang as a span, starting at the word whose sound chose the next one and ending where the sound-driven run stops, then rate severity from the count across the interview. This is instance-level coding, the approach described in evidence mode vs instance mode annotation, and it is what lets a second coder agree or disagree with a specific span rather than an impression.

A synthetic exchange with the decisions marked:

Interviewer: How did you get here today?

Speaker: The bus, the fuss, the trust, I don’t trust the driver, Clive, he’s alive, I’m alive and kicking, ticking. [Clanging: span from “bus” to “ticking”; rhyme (bus, fuss, trust), a sound-alike pun (trust to driver’s name Clive, alive), and a rhymed close (kicking, ticking); intelligibility impaired, redundant words introduced] Then I walked from the stop.

Three rules keep it honest. Mark the span at the words, not the sentence, because the rating depends on being able to point at the rhyme or pun. Check that meaning was lost: a rhyme that answers the question is not clanging. And count occurrences across the whole interview before choosing a severity, because the TLC anchors are frequency bands, not an impression of how striking the worst instance was (Andreasen, 1986).

Calibrate a coding team on the high-agreement items first (pressure of speech 0.89, derailment 0.83); clanging’s 0.58 leaves more room for disagreement, and most of that room is in the boundary with ordinary wordplay.

What are the most common mistakes when coding clang associations?

The most common mistakes are rating ordinary wordplay as clanging, missing clanging that hides inside pressured speech, and rating from one striking instance. Each has a fix:

• Rating a deliberate pun. If the thread survives and the speaker meant it, it is wit. Clanging requires impaired intelligibility and redundant words (Andreasen, 1986).
• Missing clanging inside pressure of speech. Fast, voluminous speech buries rhymes. Read the transcript slowly for sound links before rating derailment alone.
• Confusing clanging with neologisms. A rhyme on a real word is clanging; an invented word is a neologism, a separate TLC item.
• Rating from the worst moment. Severity is a count across the interview: once, two to four, five to ten, more than ten (Andreasen, 1986).
• Reading a diagnosis into the sign. Clanging leans toward mania but occurs in schizophrenia and, briefly, in people who are tired or stressed (Andreasen, 1979b; 1986).
• Treating a low kappa as coder failure. Clanging is rare; agreement on rare items is fragile in any sample. Calibrate on the common items first.

Pick the instrument for the question, not the other way round. Clang associations are the easiest thought-disorder sign to recognize and one of the hardest to rate well, and the difference is discipline: point at the rhyme, confirm the meaning was lost, and count.

References

• Andreasen, N. C. (1986). Scale for the Assessment of Thought, Language, and Communication (TLC). Schizophrenia Bulletin, 12(3), 473–482. doi:10.1093/schbul/12.3.473
• Andreasen, N. C. (1979a). Thought, language, and communication disorders. I. Clinical assessment, definition of terms, and evaluation of their reliability. Archives of General Psychiatry, 36(12), 1315–1321. doi:10.1001/archpsyc.1979.01780120045006
• Andreasen, N. C. (1979b). Thought, language, and communication disorders. II. Diagnostic significance. Archives of General Psychiatry, 36(12), 1325–1330. doi:10.1001/archpsyc.1979.01780120055007

If you code disorganized or manic speech from transcripts, Tagaroo curates the TLC thought, language and communication items and the SAPS positive thought disorder subscale as span-level annotation schemes, with the sound-or-meaning rule above built into the coding guidance and inter-rater reliability computed as your coders work. The related drift signs are covered in tangentiality vs circumstantiality.