There is a limit to what I can do with a sentence like this one:
Prior to the commencement of any maintenance activity, it is essential that personnel ensure that the appropriate lockout-tagout procedures, as outlined in Section 4.2 of the Site Safety Manual, have been fully implemented and verified.
I can slow it down. I can find the stresses, breathe in sensible places, and lift the last clause so it does not die on the vine. What I cannot do is make it sound like a person talking, because it was never built to be one. It was built to be read with the eyes, at your own pace, with the option of going back to the start.
This is the part of “conversational” that gets missed. What buyers mean by a conversational voice over covered the word and how to brief it. This one is about where the read actually comes from, which is mostly the page, not the booth.
Give me an e-learning script written for the ear and a conversational read is nearly automatic. Give me copy written for the eye and the best I can offer is a well-performed document. Here is what separates them.
1. Write to one person, in the second person
Instructional copy drifts into the third person almost by reflex. “Employees should ensure,” “users are required to,” “personnel must verify.” It feels correct and official, and it puts a pane of glass between the learner and the content.
Speech does not work that way. When you explain something to a colleague you say you.
Before: Employees should ensure that all safety equipment is inspected prior to use.
After: Check your safety equipment before you use it.
The second version is shorter, it is aimed at somebody, and it can be spoken. And it fixes a subtle problem with the first: a learner reading about “employees” is not necessarily thinking about themselves.
The related habit is the collective we for the organisation. “We take data protection seriously.” Fine occasionally, and quickly wearing. If the module is teaching somebody to do something, the pronoun that matters is you.
2. Use contractions, nearly everywhere
In ordinary speech most people contract “do not,” “cannot,” and “you will” without thinking. Writing them out in full is the single fastest way to make a read sound stiff, and it is usually a habit rather than a decision.
Before: You will not be able to submit the form if any mandatory field has not been completed.
After: You can’t submit the form until every required field is filled in.
I can contract them myself at the mic, and often do. But when I do it I am editing your script live, which introduces exactly the sort of small drift between the written and recorded versions that causes trouble later when someone updates a module and cannot find the line.
Better to write them in. If a specific line genuinely needs the full form for emphasis, that is a real choice, and worth flagging in the script so I read it as intended rather than assuming it was an oversight.
The exception is content that has to carry weight, such as safety and compliance. That is the subject of the next post, on compliance training voice over.
3. One idea per sentence
This is the big one. If you only fix one thing, fix this.
Written English tolerates long sentences because the reader controls the pace and can re-read the start. Speech has neither luxury. A listener gets one pass, in real time, and a sentence carrying three ideas across four clauses will lose them somewhere in the middle regardless of how well it is delivered.
Take the lockout-tagout sentence from the top. Here it is written for the ear:
After: Before you start any maintenance work, stop. Check that lockout-tagout is in place, and check that someone has verified it. The full procedure is in section 4.2 of the Site Safety Manual.
Three sentences instead of one. Every idea gets its own breath. And notice what that did to the delivery options: I can now land “stop” properly, which is the whole point of the paragraph and which was buried mid-clause before.
A rough test: if a sentence needs more than one breath to say, it needs to be more than one sentence.
4. Ask a question now and then
Questions are the most underused device in instructional writing, and they are free.
Before: There are three common causes of pump failure.
After: So why do these pumps fail? Usually, one of three reasons.
A question does two things at once. It gives me a genuine change of gear to play, which breaks up the metronome that long narration falls into. And it gives the learner a half-second of anticipation, which is the closest thing to attention you can manufacture in a script.
Do not overdo it. One every few screens is plenty. Used sparingly it is a hinge; used constantly it becomes a tic, and a slightly patronising one.
5. Cut the words the slide is already showing
The most common avoidable line in e-learning is the narrator reading out the heading that is on screen.
Before: (on screen: “Step 3: Configure the alerts”) “Step three. Configure the alerts. In this step, you will configure the alerts.”
After: (on screen: “Step 3: Configure the alerts”) “Now let’s set these up so the right people hear about a problem.”
The learner can read. Narration that duplicates the screen is not reinforcement. It is dead air with words in it. The voice should be doing something the slide cannot: adding the reason, the context, the caution, the thing you would lean over and say.
Same goes for “In this module you will learn,” “As you can see on the screen,” and “Let’s take a look at the following diagram.” All of them are throat-clearing. Cut them and start with the content.
6. Read the script aloud before you send it
If you do nothing else on this list, do this one.
Read your script out loud, at pace, in one pass. Not silently, not skimming. Out loud, at the speed you would actually say it.
Every place you stumble is a place I will stumble. Every sentence where you run out of breath is too long. Every phrase that makes you feel faintly ridiculous saying it is a phrase that will sound faintly ridiculous in the module. Every acronym you had to think about is one to add to the pronunciation list. The same goes for numbers, URLs, and anything you would say differently from how it is written.
It takes about as long as the finished audio, which for a ten-minute module is ten minutes. I have never seen a script that did not improve from it, and it catches a class of problem that no amount of reviewing on screen will find, because on screen you are reading with your eyes and the whole issue is that the ears are different.
What a better script saves on the voice over
This is not a style preference. It shows up in three places on the invoice and the timeline.
Runtime. On longer e-learning voice over projects, runtime is the number the cost follows. Copy written for the ear is almost always shorter than the same content written for the page, because most of what gets cut is scaffolding rather than substance. Spoken runtime also differs from written word count, so how written and spoken word counts differ is worth a look before you estimate.
Retakes. Tongue-twisters, breathless sentences, and clauses that can only be parsed on paper all produce multiple takes. Some of that lands in my time, but the ones that need a decision from you land in your inbox as a question, and each one can cost a day.
Revisions. A surprising proportion of “can we re-record this line” requests are not about the performance at all. Somebody heard the sentence out loud for the first time and realised it did not say what they meant. Reading aloud at the draft stage moves that discovery to before the recording instead of after it.
What I still do at the mic
None of this makes the read automatic, and I am not arguing myself out of a job.
Once the script is right, there is still the work: finding which word in a sentence carries the meaning, deciding where a pause does more than a comma, keeping the energy level identical across a module recorded over three weeks, pacing against on-screen motion so the voice and the cursor stay in step, and getting your product names and clinical terms right every time. That is the craft, and it is most of what you are paying for.
But it all sits on top of the sentence. The structure of a sentence decides how it can be delivered, and no amount of skill in the booth rewrites the sentence. Which is a good thing, really. It means the biggest single improvement to your finished course is available to you for free, at the draft stage, before anyone books a studio.
A conversational e-learning script checklist
Before a script comes to me, run it past these six:
- Is it addressed to you, one person, in the second person?
- Are the contractions written in?
- Does every sentence carry one idea, sayable in one breath?
- Is there a question every few screens to break the rhythm?
- Has the narration stopped repeating what is already on screen?
- Has somebody read the whole thing aloud, at pace, start to finish?
If the answer to all six is yes, the conversational read is mostly already done, and my job is to not get in its way.
Next in this series: why some modules should not be conversational at all, and how to tell which ones. Read why compliance training voice over needs a different tone.
Got a script that is nearly ready? Send it over with the word count and how the course will be used, and I will come back with a clear quote and a free audition of your actual copy. Get in touch here or call +91-99620-28747.