Why Is One Phrase Harder to Say Than Another? Machinery That Collides

Speaking is not a matter of retrieving one sound at a time, since the system prepares several upcoming sounds simultaneously in order to produce them smoothly, and that overlap is where the trouble starts. When two similar sounds are being prepared close together, the plan for one can be substituted into the position of the other, producing an error. The difficulty is therefore greatest when sounds are similar but not identical, since identical ones cannot be confused and very different ones are not competitors. Alternating patterns are worse than repeating ones for the same reason, because alternation requires the system to keep switching between two competing plans rather than settling into one.

The features that produce difficulty are specific and testable:

• •Pairs of sounds made in nearby positions in the mouth
• •Alternation between two similar sounds rather than repetition of one
• •Similar vowel and consonant frames across successive words
• •Sequences requiring the tongue to move between front and back rapidly
• •Length, since errors accumulate as working memory fills
• •Speed, since the overlap between planned sounds increases

These phrases are a standard tool in the study of speech production, because they generate errors reliably in a controlled way when spontaneous errors are far too rare to study. The errors produced follow rules rather than being random, which is the finding that makes them useful. Consonants exchange with consonants and vowels with vowels rather than crossing categories. Sounds exchange between the same position in successive syllables, so beginnings swap with beginnings. The results are overwhelmingly possible sequences in the speaker's language rather than impossible ones, which indicates that the system applies its rules even while making mistakes. Those regularities are the main evidence for how speech is planned in stages.

The phrases have practical applications that go beyond amusement. Speech therapists use them as graded exercises for articulation difficulties, selecting sequences that target the particular sounds a person is working on. Actors and broadcasters use them as warm-up routines before performance, on the reasoning that exercising the articulators at their limit makes ordinary speech easier, which is widely practised and not well tested. Language teachers use them to drill contrasts that learners find difficult, since a phrase built entirely on a distinction forces attention onto it. Researchers studying second language acquisition use them to measure how well a learner has acquired a sound. And they appear in tests of intoxication, on the reasonable basis that the ability to produce one degrades early.

Every language has these and what makes them hard reflects the sound system of that language specifically. Tonal languages produce phrases difficult because of tone sequences rather than consonants, and a famous Mandarin example consists of a long passage in which every syllable has the same sound and differs only in tone. Languages with consonant clusters produce difficulty by stacking them, which is the basis of well-known Czech and Polish examples. Languages distinguishing sounds that English does not build twisters around exactly those contrasts. Translating one is therefore impossible in principle, since the difficulty lives in the sounds rather than the meaning, which is why versions in different languages are unrelated rather than translated.

Speaking prepares several upcoming sounds at once, and similar sounds prepared close together get substituted for each other, which is why alternating patterns of nearby sounds are hardest. Errors follow rules, exchanging consonants with consonants and beginnings with beginnings, which makes these phrases a standard research tool. The difficulty lives in the sounds, so examples cannot be translated between languages.