Neologisms in Schizophrenia: Examples and the TLC Rules

Neologisms in schizophrenia are the sign everyone has heard of and almost nobody sees. A neologism, in the TLC’s definition, is “a completely new word or phrase whose derivation cannot be understood”: the TLC’s published examples are a dish thrown “at the geshinker” and a plan that got “bawked” up (Andreasen, 1986). They are vivid, they are in every textbook, and Andreasen’s data say they are “so infrequent as to be of little use in assessing most patients.”

That gap between fame and frequency is the useful thing to know, and it shapes how the sign should be coded: carefully, rarely, and with the far commoner word approximation kept firmly separate.

What is a neologism in psychiatry?

A neologism, in Andreasen’s TLC, is “a completely new word or phrase whose derivation cannot be understood” (Andreasen, 1986). The definition hinges on derivation. If a listener can work out how the word was built, from an existing word bent to a new use or from standard rules of word formation, it is not a neologism; the TLC says such forms “should be referred to as word approximations” instead.

The TLC’s own examples make the test concrete. “I got so angry I picked up a dish and threw it at the geshinker” and “So I sort of bawked the whole thing up” (Andreasen, 1986). There is no route from English to “geshinker,” and the TLC offers no derivation for “bawked” either. Clinical descriptions commonly add that the speaker uses the word fluently, as if it had always existed; that is a clinical observation rather than part of the TLC definition.

The TLC’s judgment on the item’s clinical value is unusually blunt. “Neologisms are quite uncommon,” it says in the item text, and the introduction goes further: “some TLC disorders considered to be very important in the past, such as neologisms or blocking, are in fact so infrequent as to be of little use in assessing most patients” (Andreasen, 1986). The full item set, and the reasoning behind which items earned their place, is covered in our guide to the Thought, Language and Communication scale.

What do neologisms in schizophrenia sound like?

Neologisms appear as an otherwise ordinary sentence containing one word whose derivation cannot be understood (Andreasen, 1986). The replies below are synthetic; the invented words are chosen to have no recoverable derivation.

Interviewer: How are the new tablets?

Speaker: The nurse brings the morvel at six and I take it with water. It’s smaller than the old one.

Interviewer: What did you do yesterday?

Speaker: I went down to the shops and bought a tressock for the kitchen, the old one had gone.

Interviewer: Who visits you?

Speaker: My brother, and sometimes the plindy from the council. She’s all right.

Notice three things. The grammar is intact and the sentence is otherwise meaningful, which separates neologisms from the sentence-level breakdown of word salad, covered in word salad vs derailment. The new word sits in a slot where a real word belongs (a tablet, an object, a person), so the listener knows its part of speech without knowing its meaning. And nothing about the sound of the previous word produced it, which separates neologisms from clang associations, where rhyme or pun chooses the next word.

A severity rating counts occurrences across the interview: mild for one neologism, moderate for two to four, and severe for more than five (Andreasen, 1986). Unlike the first nine TLC items, neologisms and word approximations are scored 0 to 3, with no “extreme” anchor, and the paper leaves exactly five occurrences unscored between its moderate and severe bands.

Neologism vs word approximation: can the derivation be understood?

A word approximation is an old word used in a new and unconventional way, or a new word built by the conventional rules of word formation, so that its meaning is often evident even when the usage seems peculiar; a neologism is a new word whose derivation cannot be understood (Andreasen, 1986). The TLC lists the item as “Word Approximations (Paraphasia, Metonyms),” and it excludes two things a coder will meet: paraphasias count only when formal testing for aphasia is negative, and a metaphor used knowingly is not a word approximation (Andreasen, 1986).

The TLC’s examples of word approximations include ones that are memorable precisely because they can be decoded: a ballpoint pen referred to as a “paperskate,” and a set of stock-word coinages in which a watch is a “time vessel,” the stomach a “food vessel” and a television set a “news vessel” (Andreasen, 1986). Each is strange, and each is understandable. That is the whole distinction.

Sign	Can the derivation be understood?	TLC examples	Severity bands	Weighted κ
Neologism	No	"geshinker," "bawked"	1 / 2–4 / >5	0.39
Word approximation	Yes, usually	"paperskate" (pen), "time vessel" (watch), "news vessel" (television)	1 / 2–4 / >5	−0.02
Clanging	Yes; sound chose the word	sense → cents → dollars	1 / 2–4 / 5–10 / >10	0.58
Incoherence (word salad)	The sentence itself fails	clauses joined incoherently	frequency bands	0.88

Why does the distinction matter if both score badly on reliability? Because both items carry the same severity bands, so mislabelling moves a count from one item to the other without changing the band structure, and the derivation test is the only thing that decides which item a word belongs to. A stock-word system like the “vessel” family can run through a whole interview; tagged correctly it is a severe word-approximation rating, tagged wrongly it is a severe neologism rating, and the two mean different things to a reader.

Why are neologisms rated so unreliably?

Neologisms reached weighted κ = 0.39 (unweighted 0.49) and word approximations −0.02 in Andreasen’s reliability sample of 113 patients, against 0.89 for pressure of speech and 0.83 for derailment (Andreasen, 1986). The paper does not explain the gap. One standard reading is that kappa is unstable when an item is rarely present, the prevalence problem, and a negative value means observed agreement fell below what chance alone would predict.

The practical lesson is about where to spend calibration time. Neologisms sit among the first eleven items, which Andreasen groups as “more suggestive of severe psychopathology,” yet the same introduction calls them “so infrequent as to be of little use in assessing most patients” (Andreasen, 1986). A coding team that calibrates on pressure of speech, derailment and poverty of content will agree on most of what it sees; a team that starts with neologisms will spend its time on a handful of words. Our guide to loose associations covers the common item that should come first.

How do you code neologisms from a transcript?

Code a neologism as a span over the word itself, with a note stating that the derivation could not be understood, and rate severity from the count across the interview. Instance-level coding, described in evidence mode vs instance mode annotation, matters more for this item than for any other, because the whole disagreement between two coders is usually about one word and whether it can be decoded.

A synthetic exchange with the decisions marked:

Interviewer: How do you spend the afternoons?

Speaker: I watch the news vessel until the carer comes. [Word approximation: “news vessel” for television; derivation understandable, stock-word pattern] Then we do the forms, the ones for the housing, and she checks my brennick. [Neologism: “brennick,” no recoverable derivation, used as an ordinary noun] Then tea.

Three rules keep the coding honest. Write down the derivation you tried: if you can state how the word was built, it is a word approximation, and your note is what the second coder will check. Do not promote a mishearing or a transcription error to a neologism; check the audio, because automatic transcription can render a mumble as a non-word. And treat a single neologism as a mild rating and nothing more, since one new word in an interview is exactly what the TLC’s mild anchor describes (Andreasen, 1986).

Done this way the rarest TLC item becomes a defensible one: a word, a note on its derivation, and a count. That is more than most coding schemes manage for the sign, and it is the only form in which a reviewer can check it.

What are the most common mistakes when coding neologisms?

Common coding errors include tagging word approximations as neologisms, tagging transcription errors as neologisms, and expecting the item to carry diagnostic weight it cannot bear. Each has a fix:

• Promoting a word approximation. If the derivation can be understood, the TLC says to call it a word approximation (Andreasen, 1986). “Time vessel” is not a neologism.
• Trusting the transcript over the audio. ASR and fast typists both produce non-words. Check the recording before tagging.
• Rating slang, dialect or jargon. A word that is new to the coder is not new to the language. Search before you tag.
• Reading a diagnosis into one word. The TLC’s definitions are not limited to schizophrenia, and it calls neologisms of little use in assessing most patients (Andreasen, 1986).
• Calibrating on the rare items first. Agreement is built on the frequent items; the rare ones follow once the derivation test is shared.
• Expecting an “extreme” rating. Unlike the first nine items, the bands stop at severe (more than five) for both neologisms and word approximations (Andreasen, 1986).

Pick the instrument for the question, not the other way round. Neologisms in schizophrenia are a real sign with a precise definition and a tiny base rate, and the discipline is to apply the derivation test, write it down, and resist the pull of the famous word.

If you code disorganized 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 derivation test above built into the coding guidance and inter-rater reliability computed as your coders work. The 1979 papers that precede the TLC are the definitions and reliability study (Andreasen, 1979a) and the diagnostic comparison (Andreasen, 1979b).

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