If your translation technique should be improved is only contingent on it being fit for its purpose. For instance, if it allows you to understand Japanese texts at a reasonable pace and if that is what you are going for, then it is appropriate.
For as to how you could go about improving it, I will suggest you take a look at "glossing" because what you are doing is quite close to an "interlinear gloss", which more or less means "explaining between the lines". Glossing abbreviations used by linguists, if you can understand what they mean, also help a lot when you try to translate a language that is quite different from the one you are translating to, as would be the case going from Japanese to English and might be more useful than those used in language learner books, though do draw inspiration from those.
A glossing technique
As an example, I am not a linguist and so have limited knowledge of the field, but I am using a glossing technique derived from what I saw in the constructed language communities with some added pragmatics in order to translate foreign languages. It goes more or less like this:
- Write down each sentence on a line, leaving a lot of room between each word, or even better yet, split the phrases on separate lines if you can. If you can't distinguish words, as could be the case with Japanese, leave at least five character's worth of space between each of them.
- Leave multiple blank lines between the lines from the original text. Five blank lines will do. You can go more compact, but you may sometimes run out of room.
- I do not usually do this, but for a language where word bounds are not evident, I would translate each character first or bits I know at a glance, aligning the translation with the character.
- Word by word, aligned with the words, write the meaning underneath the original word. Words that translate to multiple words are kept as one word but written with dashes. For instance, I translate the Korean word "너와" as "with-you". For any piece of information that cannot be conveyed as words in your target language (i.e. English), add glossing abbreviations to express them, separated by dots. A list of such abbreviations can be found here: https://en.m.wikipedia.org/wiki/List_of_glossing_abbreviations "わたしは", for instance, could be glossed as "I.TOP" or "me.TOP" to indicate it is the topic of the sentence, a concept I find is expressed too differently in English to translate on a word-per-word-basis, but you could use "about-me" or "on-the-topic-of-me" if that suits you better.
- Write additional potential translations underneath the previous ones. Context should help you rule some out, but most often you won't be able to rule out all of them before you get a whole sentence ready to translate.
- From the glossing of the whole sentence, try to rule out the interpretations of words that don't work. I cross them out as I eliminate them.
- Group together meanings from multiple characters or words that form more complex expressions with accolades and write the more complete idea underneath using the same syntax.
- Once you have a good understanding of each part of the sentence, try to compose a sentence that conveys that idea in the language you are translating to.
- Refine that sentence into a second, more natural one, if needed.
Example
Here's an attempt of mine at translating the first sentence of a Korean song. I can't vouch for the accuracy of the translation, but that's me actively trying to parse meaning out of it:
너와 마주 앉아 입 맞춰 부르던
with-you face-to-face to-sit mouth to-set to-sing.RET.ATT
\_____V____/ (we)-sang
to-kiss
노랫소릴 기억합니다.
singing-voice.DET? to-remember-FORM.POL
melody
I remember the melody-that-I-sang-with-you-kissing-sitting-face-to-face.
I remember the melody that we sang, kissing while sitting facing each-other.
On using a computer
I find the whole process easier to do on paper, but you could leverage software to do the same thing. A word processor would allow you to realign text much more easily, so would putting the words in cells of a spreadsheet, but some signs like the accolades would be annoying to do and you might have to fight with the auto-correct feature.
