How do I set, save and keep the same voice across all 16 parts in a pattern? What could I be doing to trigger different voices in different sections in the first place?

The Motif ES has a Phrase Header... This is responsible for memorizing what Voice was selected when you record the Phrase to a Pattern. It also stores the original Volume and original Pan position on a per Phrase basis. There are 16 potential Phrases per Pattern Track; one per Section A-P.

_ There is a Track Voice parameter - applies the same Voice to the entire Track, all 16 Pattern Sections use the same Voice.
_ There is a Phrase Voice parameter - applies the Voice on a per Phrase basis... Each Phrase on a single track could recall a different Voice. (Which is how you have it currently set)
A Phrase is any recording done in Pattern mode.

If you are unaware of these parameters it may seem like there is a rouge operating your Motif ES - I don't believe in ghosts! 🙂

Here’s how it works. The Motif ES automatically memorizes the VOICE with which you recorded the data. It’s a feature - you wouldn’t want it any other way - imagine if it did not memorize the VOICE that you used? Chaos. This stored data can be edited, you cannot simply change the Voice on the mixer, because, as mentioned the Phrase Header will change the Voice back to the stored values. Here's how to view what you've stored in the Phrase Header and how to manage this important information:

Rewrite the Phrase Header
Press [RECORD] as if you were going to record some data (because we will) but do not press PLAY
Select a Track to view [1]-[16]
Press [F2] VOICE
Here is where the Motif ES has memorized certain things about what you originally recorded to this track... notice it has documented what VOICE you used, what VOLUME, what PAN position etc.
Highlight the VOICE and change it to the one you prefer to use for this particular Phrase.
Press STOP

What you have done is record (write in) a new VOICE to play this phrase. The Motif ES now will memorize this as the sound it will use whenever you recall this USER PHRASE.

This is what is called the “PHRASE VOICE”... You can set each Track to select the "Phrase Voice" or you can set one Voice for all Sections called the "TRACK VOICE".

How to override the Phrase Header
You can have a consistent layout of instruments from Section to Section, you can set the track so that the TRACK VOICE is instructed to be recalled, as opposed to each PHRASE recalling a different Voice. By setting up to use track voice settings the mixer will remain at the most recent value without changing ... In other words, it will follow values as indicated in the track only. When PHRASE VOICE rules apply each phrase could cause fundamental changes in the mix to move around.

In pattern mode each track is either set to "Phrase Voice" On or Off, if Phrase Voice is Off then "Track Voice" applies
From the main screen press [F3] TRACK
Press [SF3] TRACK VOICE

Set PHRASE VOICE to OFF for the current Track: this will make the MIXER “live”.
The consequence is that all Phrases that are placed on this track will have the Voice you select in the MIXER.

Thanks, you resolved one of life's (or at least my ES's) greatest mysteries. Now that I understand the headers, I think I have some remaining ghost wantabes in the attic, meaning I had previously copied tracks and parts though patterns which will change once I choose my phrase vs track header. No problem, I can bust them.

May I ask one followup? - If I turn off phrase voice, meaning I have a track header configuration, does that mean that the most recent / last changes I make to numbered track (regardless of which pattern part they are connected to) will then control the voice of that track on all parts within the pattern?

the most recent changes that you have [STORED]!!! 🙂