Testing a new approach to subject markers in conversation

I spent the last week testing a "context-heavy" rule for markers. Previously, I sounded robotic by over-specifying. By dropping certain particles during fast-paced talk and only using them at logical branches in the conversation, I noticed a significant decrease in listener confusion. Has anyone else measured the impact of intentional omission versus strict textbook usage in casual settings?