In the tokenizer, spaces do not exist independently but stick to the preceding word, so 'strawberry', ' strawberry' (with a leading space) and 'Strawberry' (capitalized first letter) are three completely different token sequences.