Images in Language Space: Exploring the Suitability of Large Language Models for Vision & Language Tasks