Verbal Overshadowing is an interesting perceptual mechanism where giving a verbal description of a complex stimuli (for example: a human face) will lower the precision of the visual memory of the said thing. It has been reported to occur in connection with many different types of stimuli: images, music, colors but also smells and tastes. Verbal overshadowing has been studied for many years, with the groundbreaking study by Carmichael from 1932 that first showed that introducing a verbal description can interfere with the encoding of the visual part, allowing for many studies that followed this trail in order to uncover the applied implications.
The suspected mechanism that perpetuates this phenomenon is a discrepancy between our proficiency in creating visual memories and our ability to verbally describe things. When the pictures inside our minds often have vivid colors and great detail, we struggle to convey all of those things with words. And as it turns out, once we explicitly try to describe the images verbally, we muck up the source memory itself. After we hear our description of things, we simply cannot get rid of the influence it has on the memory, even if we have really struggled to find the right words to apply. More information about the mechanism itself: https://link.springer.com/article/10.3758/BF03201107
The biggest real-life consequence of the Verbal overshadowing is in police forensics. In general, the more research we do, the more we realize that eye-witnesses, once thought to be ironclad, are actually subjects to so many perceptual distortions and effects that anything they say should not be treated as evidence, maybe a mere guidance at best. VO is most prominent in the situation where a witness/victim has to recognize a face from a line-up. If a verbal description of a suspect is asked to be given before the actual line-up, chances that the memory of a face, once so vivd, is now a murky portrait derived from our own words.