Prosody

Also called Intonation, Stress and rhythm, Suprasegmentals, Tone of voice

The music of speech: stress, rhythm, pitch and pace. It carries the question mark, the sarcasm and the feeling, and it is what a synthetic voice loses first.

What it is

Prosody is everything about speech that isn't the individual sounds: the rise at the end of a question, the stress that turns "I didn't say that" into five different sentences depending on which word gets it, the pace that says you're excited, the flatness that says you're not. Babies respond to prosody before they understand a word, and adults read it faster than they read content.

It's also the part of speech that text-to-speech does worst. A synthetic voice says the words; it rarely says them the way you meant.

Who it can help

Understanding prosody helps anyone working with a person whose prosody is different: some autistic people have distinctive prosody (often described as flat or sing-song, and often perfectly expressive to anyone who listens properly), people with Parkinson's lose volume and variation, people with dysarthria may have an even rhythm, and people after right-hemisphere stroke may lose the ability to produce or read it.

It matters to every AAC user, because a device gives you the words and takes away the tune. Anger, tenderness, sarcasm and "I'm asking, not telling" all have to be put back some other way, and the person's face, body and choice of words do most of that work.

How it works

Prosody has four main ingredients: pitch (high or low, and its movement), loudness, duration (how long sounds are held) and pause. Together they make stress (which syllable or word stands out), intonation (the melody of the sentence) and rhythm. English is a stress-timed language: the stressed syllables land at roughly even intervals and the rest get squeezed, which is one reason English-learners with a different first language sound "foreign" even when every sound is right.

Children acquire prosody early. Babbling picks up the intonation of the home language in the first year, and toddlers use rising tone to ask before they can form a question. Later, prosody signals the difference between a list and a finish, a joke and a statement.

On a device, a person has three controls: the voice, its speed and its pitch, which the Talk board lets them set. That's a fixed setting, not a live performance, so the rest is conveyed by punctuation (a question mark changes some voices' intonation), by word choice ("seriously?"), and by the person themselves (facial expression and body language).

Getting started

Notice what the tune is doing. Read a sentence to a child three ways: as a question, as a surprise, as a bored statement. Ask which was which. Play with it.

For a speaker with reduced prosody, don't ask them to "put more expression in"; it doesn't work. Work on loudness first (it's the most trainable), then on pausing at the ends of phrases, which makes speech easier to follow even with no melody.

For an AAC user, set the voice deliberately, with the person, and try the speed and pitch controls until it sounds like them (choosing a voice, voice speed and pitch). Then put emotion words and emphasis words on the board: "really", "seriously", "joking", "I mean it". Those are the person's prosody now.

Where to get help

The Royal College of Speech and Language Therapists for the profession that treats prosody problems. Parkinson's UK on voice and speech changes. For AAC voices, Communication Matters on choosing and personalising a voice, and voice banking for people who may lose theirs.