T-test puzzlement

[nom]

I've been pondering on this for some time and I guess i'm confused because there are so many stats tests out there.

Basically my aim is to figure out if there is any statistical relevance between years of English learnt and correct answers on an English test.

I have put years into months for ease i.e. 1 yr 6months= 18 months.
Both are equal sets of data (20)

For marks, each is out of 12, however I have just listed 6, 9, 10 etc. just wondering if this is the right way to go? Or would a percentage be better?

Now this is the confusing part, I assumed I was going to use an equal sample size and equal variance t-test, however after realising that standard deviations and means are TOTALLY different like std dev for one is 3.4 and the other 45.9 (these are just random values btw not actual results) I guessed that this is not the way to go.
My current assumption is to use Welch's t-test, but I would just like to know if I am on the right path or not. Any guidance will be appreciated! Thanks!!

[Dason]

How many different values for the predictor (years) do you have? It sounds like you only have two groups - is this correct?

[nom]

Yep I have two sets of data. One for years, the other for correct marks.

I have 20 different values for each data. I wonder if I could put the excel file up on here to show you...

I changed the data for years and months to decimals i.e. 1year 6 months= 1.5 and so on. I also changed the correct marks from out of 12 i.e. 6/12 = 0.5 I was wondering if this is right way to go or not or to just state it as 6, 7, 11 etc.