Initialism and acronym
An initialism is spelled out letter by letter (BBC), an acronym is read as a word (NATO). Nothing in the writing tells you which is which.
An initialism is an abbreviation whose letters are spelled out: BBC is said 'bee-bee-cee'. An acronym is an abbreviation pronounced as an ordinary word: NATO is said 'nay-toh', not 'en-ay-tee-oh'. Both are written exactly the same way, in capitals, with nothing to tell them apart.
Why this is a reading problem, not a grammar one
Because a reading system has to choose, and nothing in the text helps it. Applying a single rule produces errors both ways: always spelling out gives 'en-ay-tee-oh' for NATO; always pronouncing turns a French rail operator into a nonsense word. The second error is the more frequent and the more damaging: it is audible in a second, and it casts doubt on everything read afterwards.
French news copy is an extreme case. A council report can line up EPCI, PLU, CCAS, ZAC, DGF, a mix of administrative abbreviations, some pronounceable and some not, many existing only in a local context.
The three decisions, article by article
Spell or pronounce, first, which calls for a reference list rather than a guess. Expand or not, next: the first mention is expanded if the readership does not know the term, and never afterwards, on pain of an unbearable reading. Agreement, finally: a spelled-out abbreviation behaves differently in the sentence depending on how it is read.
What this tells you about a vendor
It is a good test, because it cannot be gamed. Give a tool a paragraph containing a spelled-out initialism, a pronounceable acronym and a rare local abbreviation, and listen. A product that handles all three has done the normalization work; a product that misses one left it to the voice engine. We publish our own test cases, successes and failures alike, precisely so that nobody has to take our word for it.
Related terms
- Text normalization : The step that turns written text into speakable text: numbers become words, acronyms are expanded or spelled out, abbreviations are written in full.
- Text-to-speech (TTS) : Text-to-speech converts written text into spoken audio. Recent neural models produce voices that are hard to tell apart from human speech.
- Prosody : Everything in speech that is not the words themselves: rhythm, pauses, melody, intensity. It is what makes a reading sound human.
Give your articles a voice
WeDispatch automatically turns your articles into an audio version, the moment you publish. Try it free on your own articles: no credit card needed.
Book a demo