The standard method for ADR is to replay the video scene back to the actor on a monitor while they try to sync their replacement dialogue with the original recording. However this still somewhat hit and miss because the actor is almost always either lagging or leading their original dialog.

Why? Because they cannot exactly see the precise moment when each word begins until it has begun, at which point they are often late. Alternatively they try to second guess the timing and pre-empt themselves.

Instead of playing back their original scene as a cue give this a try.

Let the actor view the audio waveform of their original dialog on a NLE video monitor... a bigger monitor is obviously better than smaller one.

Minimise the amount of dialog on the screen at any one time to ensure the waveform of each individual word/phrase is clearly defined on the screen. If there is a lot of background noise contaminating the dialog then either remove it or at least minimise it so the individual waveforms for each word are clearly visible with well defined starts and endings.

Let them watch the screen while you loop the section of dialog several times over. Point to the cursor as it moves horizontally across the screen so they can see as the cursor hits the start of each section of waveform representing each spoken word of the orignal recording.

As they watch the cursor cycle across the screen they will quickly identify which section of waveform corresponds to which word they're hearing and thus deliver each word the moment the cursor just reaches the start of each section of the corresponding waveform.

With a little practice they should now find it easier to achieve fairly accurate sync and they can then concentrate more on the delivery and emotion.

Try it yourselves. You should find it is much easier to achieve accurate sync than by simply watching an onscreen video of the original take.

Of course capturing good sound on location is always preferable.