A key area of work, and indeed of challenges, in Machine Translation (MT) is dealing with ambiguity. While in MT we aim to reduce the gap between human perception and machine solution, we have repeatedly failed to show perfect results regarding this, and we continue to failing getting human like translations from our machines. This conceptual paper reviews a few of the commonly faced problems in ambiguity resolution when translating from English to Bengali and vice versa, and, on the basis of this, aims to propose a soft-computing based method in designing an English to Bengali and vice versa translator which may yield better results in the future. The implementation and hence the corresponding analytical studies thereafter are left as future work.