The term "flattening" is not standard, but if the book chooses to uses, that's fine. (A mathematician would probably use the term singular matrix.) For a 2×2 matrix, there are various equivalent ways of defining this property. If you have met determinants, the simplest definition is probably to say that a matrix is flattening if its determinant is 0. Otherwise, you can use the definition in the book, that one column should be a scalar multiple of the other. In the case of the matrix , each element in the left column is 2/3rds of the corresponding element in the right column. Another equivalent definition is that one row should be a scalar multiple of the other. In the case of that matrix, the bottom row is twice the top row.