Practice Punctuation and Literal Words in Dragon on a VOX Laptop
The word colon can describe a punctuation mark, or it can be a word that belongs in the text. Quote that can be a spoken Dragon command, or those words can be part of testimony about a command. A spelled surname can need exact letters while the sentence around it still needs normal punctuation. These are small distinctions, but they matter in a voice-writing practice document.
A useful rehearsal does more than ask whether a mark appeared. It preserves what the reporter intended to say, what Dragon produced, what command or formatting rule may have acted, and what reached the CAT or practice document. The result is an evidence sheet that separates punctuation, literal words, commands, spelling, and downstream formatting.
This guide uses a fictional exhibit and invented people, names, file labels, settings, and observations. Nothing in the worked example is a customer result, a court record, a performance claim, or a promise that Dragon will produce the same output on another system.
The practice goal
The goal is to answer five narrow questions:
- When the reporter says a punctuation name, does Dragon insert the intended mark?
- When the punctuation name belongs in the sentence as an ordinary word, what supported method produces that word?
- When a phrase resembles a voice command, does it insert text or perform an action?
- When the reporter spells an invented name, does the saved text preserve the intended letters and capitalization?
- Does the text change again when it moves into a CAT or other destination window?
Do not turn the exercise into a general recognition score. The passage is intentionally unusual. It tests routing and formatting decisions that ordinary prose may not expose.
Record the exact Dragon environment first
Open the installed Dragon product and record its customer-facing name, edition, version, and build. Also record the active user profile, language, microphone or stenomask, and the destination application.
| Field | Illustrative entry |
|---|---|
| Windows | Windows 11 Pro |
| Computer | Martel VOX Elite voice-writing laptop |
| Dragon product | Record the exact installed name |
| Dragon version/build | Copy from the About screen |
| User profile | Mara-Practice, fictional |
| Language/dialect | Record the active profile setting |
| Input device | Record the selected stenomask or microphone |
| First destination | Plain supported practice document |
| Second destination | Authorized CAT practice job |
| Automatic punctuation | On, off, or unresolved |
| Practice date | 2026-11-19, illustrative |
Nuance’s Dragon help explains that punctuation can be spoken at the appropriate place and that an option can automatically add commas and periods. That option changes the meaning of a comparison. If the reporter does not record it, a missing spoken command and an automatic formatting decision can be confused.
Do not assume that the visible Windows 11 input is the active Dragon source. Confirm the source in Dragon and, when the CAT application has its own audio or input selection, record that separately.
Keep five layers separate
The exercise becomes clearer when every observation is assigned to a layer.
1. Spoken plan
The spoken plan is the exact wording the reporter intends to say. It includes punctuation names, command-like phrases, spelling, and pauses. Write it before speaking.
2. Captured speech
The captured speech is the available practice audio or playback evidence. It helps determine whether the reporter actually said the planned words. Follow the application and organization rules for recording, storage, and confidentiality.
3. Dragon recognition and commands
Dragon can recognize ordinary dictated words, punctuation names, and voice commands. A command can change existing text or the application rather than insert its spoken words. Preserve the first visible Dragon result before correcting it.
4. Destination application
The destination may apply its own behavior. Focus, unsupported fields, a Dictation Box, keyboard shortcuts, or CAT integration can affect where text appears and how it is formatted.
5. CAT translation and editing
The CAT environment may translate, label, punctuate, or format the received text. That downstream result is not proof that Dragon produced the same string initially.
A single final transcript cannot reveal all five layers. Save evidence at the earliest practical point.
Use the official punctuation names for the installed edition
Current Dragon documentation lists spoken forms for punctuation and symbols. Common examples include comma, period, question mark, colon, semicolon, open quote, close quote, hyphen, and section sign. Dialect and edition can affect available forms.
Use the Help and Vocabulary Editor that match the installed product. If a spoken form is not documented there, mark it Unverified rather than borrowing a command from a different product or older release.
There is no need to assume a universal spoken command that forces a punctuation name to appear literally. When the intended text is the word colon rather than the mark :, use a supported correction, spelling, vocabulary, or word-property method confirmed for the installed Dragon edition. Record which method was used. A manual correction in the editable practice document is also a clear fallback, provided it is labeled as editing rather than as the initial recognition result.
Why “literal word” needs a definition
In this exercise, a literal word means the ordinary printed word that the reporter wants to see. It does not refer to a scripting term, an application programming interface, or a hidden recognition result.
For example:
- Intended punctuation:
Exhibit code: ZL-47 - Intended word:
The witness said the word colon. - Intended quoted command phrase:
The trainer wrote “quote that” on the card.
The first line needs a colon mark. The second needs the letters c-o-l-o-n. The third needs the words quote that enclosed by quotation marks, not the action associated with a Dragon command.
Write the intended visible text before recording the spoken plan. That target helps the reviewer distinguish a command action from text recognition or a formatting choice.
The fictional exhibit
The practice matter is Meridian Glass Review, and the fictional exhibit is ZL-47. The invented witness is Avira Noll. The exercise uses the following intended visible text:
Exhibit code: ZL-47. Avira Noll read the label “north colon line.” The trainer wrote “quote that” on a separate card. The witness spelled Noll, N-O-L-L, and then said, “Is the mark a colon?”
Everything in that paragraph is illustrative. It is designed to place a colon mark, the word colon, quoted command-like words, a spelled name, a hyphenated identifier, commas, quotation marks, a question mark, and ordinary periods in one short exhibit description.
Build a token map before speaking
Create one row for every item that could be confused with another layer.
| Token | Intended visible result | Token type | Risk to observe |
|---|---|---|---|
| T01 | Exhibit code: |
Text plus colon mark | Colon may be omitted or become a word |
| T02 | ZL-47 |
Identifier | Letters, digits, or hyphen may differ |
| T03 | Avira Noll |
Invented proper name | Spelling or capitalization may differ |
| T04 | “north colon line” |
Quoted phrase containing literal word | Colon may become : |
| T05 | “quote that” |
Quoted command-like phrase | Dragon may execute rather than insert words |
| T06 | N-O-L-L |
Spelled name | Spaces or hyphens may differ |
| T07 | “Is the mark a colon?” |
Quoted question with literal word | Quote and question mark routing |
Review each mapped token separately. The token map shows which distinction each part tests and gives each observation a clear place in the practice record.