Short answer
In normally sighted people visual thinking is the dominant mode of thinking and may be intimately associated even with verbal thinking and hence difficult to root out completely.
Background
Human thought can be generally divided into visual and verbal thinking. Visual thinking is mediated by visual imagery, where imagery visual representations are created of things that have been, are, or will be (e.g., life events such as memories of a recent vacation). In contrast, verbal thinking involves thinking in words, for example when you are thinking about what and how you will tell something to friends or colleagues, i.e., inner speech (source: Harvard Gazette, May 2017).
A study by Amit et al (2017) investigated subjects' brain activity when recalling images or sentences. Verbal thoughts resulted in more robust activation of language centers in the brain than during visualizations. However, subjects generated visual images regardless of whether their intent was to visualize or to think verbally. One possible interpretation the authors provide is that visual thinking is a primary mode of thought, which is supported by the notion that verbal abilities appeared later in human evolution and also appears later in childhood.
An interesting notion here is of congenitally blind people that lack visual memory altogether. In these instances, visual imagery is totally absent and in these people verbal thinking may be the only mode of thought available (Zimler & Keenan, 1983). In congenitally deaf-blind individuals the mode of thinking may be further warped into tactile thoughts. These considerations may lead to believe that with sufficient practice, pure-verbal thinking may at least theoretically be possible in the sighted.
References
- Amit et al., Neuroimage (2017); 15(152): 619-27
- Zimler & Keenan, J Exp Psychol Learn Mem Cogn (1983); 9(2): 269-82