Pressured Speech vs Flight of Ideas: Examples and Coding

Pressured speech vs flight of ideas is a distinction between how much someone says and how their ideas connect. Pressured speech is an increase in the amount and speed of spontaneous speech, with the speaker hard to interrupt. Flight of ideas is derailment at speed: ideas slipping off the track onto obliquely related or unrelated ones, in the context of pressured speech (Andreasen, 1986). The two often arrive together in mania, which is why they get merged, but they are coded separately, and either can occur without the other.

The cleanest way to hold them apart is Andreasen’s own decision when she built the Scale for the Assessment of Thought, Language, and Communication (TLC): pressure of speech got its own item, and flight of ideas did not.

What is the difference between pressured speech and flight of ideas?

The difference is the dimension being described. The TLC defines pressure of speech as an increase in the amount of spontaneous speech compared with what is ordinary or socially customary, in which the patient talks rapidly and is difficult to interrupt (Andreasen, 1986). Nothing in that definition is about the content. The TLC adds that pressure “may be accompanied by derailment, tangentiality, or incoherence, but it is distinct from them,” and that if rate is measured, “a rate greater than 150 words/minute is usually considered rapid or pressured” (Andreasen, 1986). Flight of ideas describes the content: ideas that slip off the track onto another that is clearly but obliquely related, or onto one that is completely unrelated, which the TLC codes as derailment. Put the two dimensions together and you get the manic pattern: a lot of fast speech whose ideas keep jumping.

Andreasen’s reason for not giving flight of ideas its own item is worth quoting, because it settles most arguments about terminology. Less severe derailments “have sometimes been referred to in the past as tangentiality or as flight of ideas when in the context of mania,” she writes, but “such distinctions are not recommended because they tend to be unreliable. Flight of ideas is a derailment which occurs rapidly in the context of pressured speech” (Andreasen, 1986). Elsewhere in the same paper the term is said to have been “dropped and is now subsumed under the concept of derailment.”

So a rater using the TLC never writes “flight of ideas” on the sheet. They rate pressure of speech on one line and derailment on another, and the combination is the thing older textbooks called flight of ideas. The full glossary is covered in our guide to the Thought, Language and Communication scale.

What do pressured speech examples sound like?

Pressured speech sounds like too much speech, too fast, that does not leave room for the other person. The TLC’s anchors are built around how long simple questions take to answer and whether the interviewer can get in (Andreasen, 1986). The examples below are synthetic.

Interviewer: How did you sleep last night?

Speaker: Fine, great, four hours maybe three, I don’t need it right now, I was up at five and I’d already answered all my emails by six and then I went for a run, which I haven’t done in years but I felt like it, and I’ve been thinking I should enter a race, there’s one in March, I looked it up, the registration is open, I could do it, I’m sure I could do it, and then I started on the garage, which, you should see the garage. [the interviewer tries twice to speak and is talked over]

Note what is not wrong with that reply. Every idea connects to the last: sleep, early waking, emails, the run, the race, the garage. A reader can follow it. If this pattern held across the interview, it would sit at moderate to severe on the TLC anchors (several minutes per simple question, difficult to interrupt), with no derailment at all.

The TLC severity anchors for pressure of speech are concrete enough to apply from a recording. Slight is some increase in amount, speed or loudness. Moderate is usually taking several minutes to answer simple questions, possibly talking when no one is listening, and speaking loudly and rapidly. Severe is frequently taking as much as three minutes to answer simple questions, sometimes beginning to talk without social stimulation, and being difficult to interrupt. Extreme is talking almost continually, being impossible to interrupt, and possibly shouting to drown out others (Andreasen, 1986).

What do flight of ideas examples sound like?

Flight of ideas sounds like pressured speech whose ideas do not stay on the track. The same question, with derailment added:

Interviewer: How did you sleep last night?

Speaker: Sleep is for people with nothing to do, I have a lot to do, the council has a lot to answer for, the bins, have you seen the bins, green is the colour of money and I’m going to make a lot of it, my brother never understood money, he’s in Sheffield, steel city, I could cut through steel today, honestly, ask anyone.

Here the pace and quantity are the same as before, but the connections are oblique or absent. Sleep to the council to the bins is a stretch; bins to green to money is a slip on a word; money to the brother to Sheffield to steel to cutting through steel is a chain of associations rather than an argument. Each slip is a derailment, and a rater tallies them: the TLC rates derailment from mild (one clear instance) to extreme by frequency across the interview (Andreasen, 1986).

A third pattern belongs in the same family. Distractible speech is stopping mid-sentence to respond to a nearby stimulus, the interviewer’s tie, an object on the desk, and it has its own TLC item (Andreasen, 1986). In mania it often travels with pressure and derailment. Unlike derailment, the trigger is external and visible, which makes it easy to code from video and nearly impossible to code from an audio transcript alone.

How does the TLC separate pressure of speech, derailment and distractible speech?

The TLC separates them by what is being measured: quantity and pace for pressure, the connection between ideas for derailment, and an external trigger for distractible speech (Andreasen, 1986). The table lines up the four items most often bundled as “manic speech.”

TLC item	Dimension	Trigger	Synthetic cue	Weighted κ
Pressure of speech	Quantity and pace	None needed	Several minutes to answer a simple question; talks over the interviewer	0.89
Derailment (flight of ideas)	Connection between ideas	The speaker's own train of thought	Bins → green → money → brother → Sheffield → steel	0.83
Distractible speech	Attention	A nearby stimulus	"…and then I, is that a new clock?"	0.78
Clanging	Sound over sense	A word's sound	"Steel city, I could cut through steel, I feel like a seal"	0.58

Pressure of speech is the highest-scoring TLC item by weighted kappa (0.89, close to incoherence at 0.88). Differences this small, in a single sample of 113 with no confidence intervals reported, are suggestive only; a plausible reason is that rate and amount are countable, while “obliquely related” needs judgment.

Weighted inter-rater kappa for six TLC items in Andreasen’s reliability sample of 113 patients (Andreasen, 1986). Weighted kappa varies by item; the paper reports no confidence intervals.

Is flight of ideas the same as loose associations?

In the TLC they are the same item: derailment is headed “Derailment (Loose Associations, Flight of Ideas),” and flight of ideas is the rapid form, occurring “in the context of pressured speech” (Andreasen, 1986). Loose associations is the older, broader name for ideas slipping between one another; flight of ideas names the same slippage when it comes fast and under pressure. In both cases the rater codes derailment and lets the pressure item carry the pace. For the related pair of drift signs that are defined by a question rather than by spontaneous speech, see tangentiality vs circumstantiality.

Which conditions show pressured speech and flight of ideas?

Pressured speech leans toward mania, flight of ideas (derailment) does not separate mania from schizophrenia, and neither is specific to either. In Andreasen’s earlier comparison of patients with mania, depression or schizophrenia, pressure of speech, distractible speech, clanging and circumstantiality were more strongly associated with mania, while derailment itself occurred at similar rates in mania and schizophrenia (Andreasen, 1979b). That last point matters for the flight-of-ideas label: the slippage is shared across the two conditions; it is the pace and quantity that tilt toward mania.

The TLC paper is also explicit that these behaviors “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). Stimulants, caffeine, anxiety, and plain excitement produce increased speech. A rater describes the sign; the clinician with the whole picture decides what it means. Where mania itself is being rated, pressured speech is one of the eleven items of the Young Mania Rating Scale, where speech rate and amount is a double-weighted item (Young et al., 1978).

How do you code pressured speech and flight of ideas from a transcript?

Code pressure of speech as a whole-interview rating and derailment as tagged spans. Pressure is a property of the interview: turn length, how often the interviewer’s attempts to speak are overridden, and the pace, which a transcript carries only partly (word counts per minute survive; loudness does not). Derailment is a property of specific moments, so each slip is marked as a span starting where the idea leaves the track, which is the instance-level approach described in evidence mode vs instance mode annotation.

A synthetic exchange with the decisions marked:

Interviewer: What’s been keeping you busy this week?

Speaker: Everything, I’ve got three projects going, the app, the book, the thing with Dev, Dev’s a genius, geniuses don’t sleep, Edison didn’t sleep, light bulbs, I’ve been changing all the bulbs in the house to the warm ones because the cold ones are a conspiracy, honestly, ask anyone, anyway the app is nearly done.

[Pressure of speech, evidence only: a simple question answered at length and pace with no pause for the interviewer; it counts toward the interview-level rating but does not set it. Derailment ×2: “geniuses don’t sleep, Edison didn’t sleep, light bulbs” (an oblique slip from Dev to Edison to bulbs) and “cold ones are a conspiracy” (an unrelated jump), each tagged as a span. The reply does loop back (“the app is nearly done”), which keeps this short of loss of goal.]

Three rules keep this honest. Rate pressure once, from the whole interview, using the TLC’s time-based anchors rather than a single long turn. Tag every derailment at the exact words where the slip happens, so a second coder can agree or disagree with a specific span. And never write “flight of ideas” as a code; write pressure of speech plus derailment, which is what the instrument measures and what the reliability data support (Andreasen, 1986). The same endpoint-style discipline applies to the pair most often confused with these, covered in tangentiality vs circumstantiality.

What are the most common mistakes when coding pressured speech vs flight of ideas?

The most common mistakes are writing “flight of ideas” as a single code, inferring derailment from speed alone, and rating pressure from one long turn. Each has a fix:

• Using “flight of ideas” as a code. The TLC dropped it. Rate pressure of speech and derailment separately; the combination is the old label (Andreasen, 1986).
• Inferring derailment from pace. Fast speech with connected ideas is pressure without derailment. Read the connections, not the speed.
• Rating pressure from one turn. One long reply is not pressure of speech. The TLC anchors are about how the interview as a whole went: minutes per simple question, interruptibility, talking without stimulation (Andreasen, 1986).
• Coding distractible speech from audio. Its trigger is external and usually visual. From a transcript alone, code it only when the speaker names the stimulus.
• Reading a condition into one sign. Pressure leans toward mania; derailment does not separate mania from schizophrenia (Andreasen, 1979b). Describe the speech and stop there.
• Treating low kappa on clanging as coder failure. Agreement on infrequent items is fragile in any sample. Budget calibration on the common items.

Pick the instrument for the question, not the other way round. In pressured speech vs flight of ideas, if the question is how much and how fast, that is pressure of speech, and two raters will agree on it almost every time. If the question is whether the ideas held together, that is derailment, and it is worth tagging exactly where they came apart.

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
• Young, R. C., Biggs, J. T., Ziegler, V. E., & Meyer, D. A. (1978). A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry, 133, 429–435. doi:10.1192/bjp.133.5.429

If you code disorganized or manic speech from transcripts, Tagaroo curates the TLC thought, language and communication items, the SAPS positive thought disorder subscale and the Young Mania Rating Scale as annotation schemes, with span-level derailment tagging and inter-rater reliability computed as your coders work.