There is no difference in meaning. There is a difference in use.
Relative clauses—the sort of clause you use, “which is blue” / “that is blue”, which tells us something more about the noun referred to by which or that—are of two sorts: restrictive and nonrestrictive.
A restrictive clause restricts the noun it modifies to what’s defined in the clause. The clause identifies the noun and is essential to your meaning. For instance:
I looked at the books which he sent me last week. … The books I’m talking about are the ones which he sent me last week.
A nonrestrictive clause adds information about the noun it modifies. The clause is almost parenthetical, it could be left out without changing your meaning:
I looked at the books, which he sent me last week. … The books have already been identified in our discourse, I’m just throwing in a by-the-way comment about when I got them.
Note the comma in that sentence: it sets the clause off and ‘marks’ it as something added.
Now Which/That …
That may only be used at the head of a restrictive clause. It is not used, in any register, with a nonrestrictive clause.
Which may be used at the head of either sort of clause.
You may encounter another rule, which is loudly disputed. About a century ago the Fowler brothers suggested a “division of labour”—using that only at the head of restrictive clauses and which only at the head of nonrestrictive clauses. This proposal made sense to many people, it was picked up by several prestigious grammarians and style guides, and in consequence it’s often cited as a “rule”.
But the fact is, this division has never been generally adopted, and there’s no reason to follow it. I myself don’t follow it; quite the opposite, I employ “which” wherever I can, because I believe that that has entirely too much work to do already and a multiplicity of thats is likely to confuse the reader.
But that’s my choice. You’re free to follow your own rule, as long as you don’t put that at the head of a non-restrictive clause.
Best Answer
Sentences
Imagine a long piece of writing. If you look at the grammar, you will see that everything is organised into 'chunks'. So if we look at the very, very small chunks, we have different parts of words that have different meanings. Look at this word:
We can break this word into three bits. We have re which means again. We have play. And we have that bit at the end ed -which means something like past. These bits, or 'chunks', are called morphemes.
Now, as we put the morphemes together we get words. We can then put the words together in bigger chunks:
There are four words here, but they are one 'chunk'. Here they are a Noun Phrase. We can put these bigger chunks together and we get sentences. In grammar, sentences are the largest chunk you have. A sentence is the maximum chunk. When you finish one sentence, you have to start another. Grammar is really about how all these different chunks work. Grammar is about what happens inside sentences.
Intonational Phrases
But - languages are not only about grammar. Languages are also about sound. Now the sound system of a language is not the same as the grammar system. Sometimes they work together - but the sound system is a different system.
In the grammar, the biggest chunk is a sentence. But in the sound system, the biggest chunk is a tune, or song. These tunes in languages are usually called Intonational Phrases. However, because Intonational Phrase is a very long name, we usually just say IP.
One important fact to know is that we don't have one tune or IP for each sentence. We might have one sentence but two or three IPs. Look at this sentence:
We could say this in one big IP. Or, we could break it up into small tunes, small IPs:
Here we can see three different IPs. Sometimes we have one sentence and one IP, but often we have one sentence and more than one tune - more than one IP.
Relative clauses.
Restrictive relative clauses make the noun behave like one big noun with lots of words. The clause effectively behaves like an extention of the noun. Non-restrictive clauses, on the other hand, feel like a little extra comment about the noun or Noun Phrase. They feel kind of separate from the noun itself.
In the first example, we can say the whole sentence in one IP. There will be a special musical stress on the first syllable of the word doctors, because it is the last new information word. Every IP, in fact, has a special musical stress. It is called the nucleus. The nucleus is usually the last stressed word in the IP. As I said, we could say the first sentence in just one IP. However, it would be extremely unusual for this to happen. Relative clauses at the end of a sentence are likely to be marked off by a separate IP, whether they are restrictive clauses or not! We will probably say the sentence as shown below:
Non-restrictive clauses nearly always have their own IP, whether they are at the end of the sentence or not. In the second example, therefore, we are also extremely likely to have two IPs:
However - note that I included the words 'nearly always' and 'usually' there. It is perfectly possible to say the sentence above in one IP. This is probably not what we wanted to hear. It would be nice, if we had a cut and dry method for distinguishing between the two sentences in terms of the number of IPs. However, in actual fact both sentences are nearly always going to be said with two IPs, and the unlikely case of there being only one IP is possible in both cases.
Notice as well, that I said nothing about any pauses between the clauses. There is not likely to be a pause after sons in example (2). Sure, we could put a pause in there, but in actual fact, it's rather unlikely. It is most likely to happen if the relative clause is an afterthought.
Interpolated relative clauses
The relative clauses in (1) and (2) occur at the end of the sentence. This is because they are modifying the object - which, of course, occurs at the end of the sentence. If they modify the subject, however, everything will be a bit different. Consider the examples below:
The four girls who were doctors lived in Paris.
The four girls, who were doctors, lived in Paris.
Sentence (3) is likely to have two IPs as shown below:
Notice that the first IP ends at the end of the relative clause. The nucleus there will be on the word doctors. The second IP carries on from there to the end of the sentence. The second special musical stress will be on the first syllable of Paris. Compare this with the probable pronunciation of sentence (4) below:
Here, the nonrestrictive clause has its own IP that clearly separates it from the rest of the sentence. This splits off the four girls at the beginning into a separate IP. We now have three nucleuses, on girls, doctors and Paris. These mark out three separate IPs. There will not necessarily be a big pause after girls here, even though it is in its own IP. Something rather interesting happens though with the relative clause. The pitch will usually drop considerably so that the relative clause is said at a much lower pitch than the rest of the sentence. This makes it clearly separate in some way from words surrounding it. However, this will not happen in sentences (1) or (2)!
Sentences 1 and 2
Even though sentences (1) and (2) are both likely to be said with two IPs, even though there is unlikely to be a pause after sons in both cases, and even though the non-restrictive relative clause in (2) will not be said at a lower pitch, there is still one thing that may help us to tell them apart.
De-accenting
Have a look at the sentences below. I have marked out the nucleuses in bold:
In the first sentence by Speaker A, there are two IPs. The words coffee and sugars take the nucleus as they're the last information words in the IP. However, when Speaker B gives their order, we can see that the words coffee and sugar have lost their nucleuses. The reason for this is that coffee and sugar are now old ideas in the sentence - we've already been talking about coffee and sugar. We don't usually put the nucleus on words representing old ideas in the conversation. This means that white and no now have the nucleus. This is sometimes called contrastive stress. Here the words white and no are contrasting with the ideas black and two.
If we have another look at sentences (1) and (2), it's quite easy to make a theory about the different reasons why someone might say sentence (1) instead of sentence (2).
Sentence (2) gives us the idea that the man had exactly four sons and that these sons became doctors. The nucleuses in the two IP's are going to be on sons and doctors. Sentence (1), however, gives us the impression that maybe there were more sons, but that four of the sons were doctors. Here the word four is very likely to take contrastive stress. The speaker is probably contrasting the number of sons who were doctors compared to the total number of sons. Sons is quite likely to be old information here - the speaker may already have been talking about the man's sons. Even if they haven't though, sons is likely to be deaccented - it probably won't take the nucleus. The nucleus is likely to be on four. Portraying the information this way implies that there are other sons in the background.
This means, then, that we are likely to be able to tell whether the sentence includes a restrictive or non-restrictive clause because of where the nucleus is placed (the tonicity in the sentence), and not because of the number of IP's ( - the tonality). The sentences are likely to be said as shown below. Again, I have marked the nucleus of each IP in bold:
I hope this helps!