The translation is wrong. Don't use Google translate for anything except:
1. getting a very vague idea of what a text says in your native language
2. having a laugh
Any dictionary (e.g. try searching for WWWJDIC) should tell you what kanji can be used for a word if you input the kana (often more than one option exists), but sometimes the kanji are actually not used that often.
I checked out of curiosity - kanji doesn't make it any better, although switching out ちっちょい for something like ちいさい improves the translation a lot. Google use a statistical algorithm, which is why it can fall over like this even when the structure is something simple like AけどB.
Statistical algorithms use large bodies of parallel text to train on, which tend to be heavily weighted towards written and formal language - e.g. Canadian parliament records are available in both English and French, the EU put out official documents in multiple European languages, Hong Kong puts out documents in both English and Chinese.
That's why they don't work so well on phrases containing words like ちっちょい which probably don't appear that often in the input texts.
(I was just reading a paper where they searched for webpages that had multiple language versions and used those as parallel texts to feed a search engine translator - e.g. search "architecture in Berlin" in English, get returned appropriate pages in German with the option to translate them back to English. Things like this are harder than people think because computers are much stupider than people think).
Sorry for the digression - machine translation is an interest of mine.