The required has a word length of 16 bits and we are given chips of word length 8 bits, hence we would need horizontal stacking of 2 chips to get the word to 16 bits.
The total capacity needed = 2048 * 16 bits
The capacity of one chip given to us = 128 * 8 bits
Therefore the total number of chips will be = $2048 * 16 / 128 * 8 = 32$ chips
Now as we know that there will be two chips side by side to make the word length 16 bits the number of chips in vertical fashion will be = $ 32/2 = 16$ chips
The data line will be of 16 bits since word size is 16 bits
For address, we will need $4 bits$ to select the chip number (which one out of 16 to select from vertical alignment) and $7 bits$ then to choose from the 128 words present in that individual $128X8$ chip.
The diagram will look as follows: