As far as I can tell, the only difference between these two ways of describing classes of words is that 'syntactic categories' actually relies on evidence of use for determining categories, while 'parts of speech' labels can just be applied without any appeal to evidence. There is also the term 'word classes' which appears to be similar, but also used without specification. Are they simply used by different sub-disciplines of linguistics? Or are there more specific differences?
The term part of speech is narrower than the term syntactic category. Every part of speech is a syntactic category, but many syntactic categories are not parts of speech. This holds particularly of the distinction between word and phrase. Words are parts of speech, but phrases are not. A noun is a part of speech and a syntactic category, whereas a noun phrase is a syntactic category but not a part of speech. An adjective is a part of speech and a syntactic category, whereas an adjective phrase is a syntactic category but not a part of speech. Thus the term part of speech denotes a subgroup inside the greater group denoted by syntactic category.
I hazard the following answer. If you will consider for a moment, parts of speech are word categories, and can be used to create phrases. So they are essentially phrase-level syntactic categories, functional modifications necessary to make correct phrases. The concept of "syntactic category," if I am not mistaken, includes these plus sentence level concepts like subject, object, predicate, clause, etc. A noun is a word root so modified morphologically or simply by word position as to be capable of heading a noun phrase, combine with adjectives, or other functional roles; but a NP as such has no "subject" nor is one in itself. "Subject" is a sentence-level functional category. In the modern sense this a syntactic category. Of course in languages like Latin sentence level morphology is built into words through inflection. If you take the phrase puella bona you have an item morphologically marked for functional role as sentence subject. But it is of course still only a noun phrase and not an actual subject. And puella can never head a VP or PP, not because of its inflection, but because it is a noun.
What follows is just my impression.
Syntactic category is a term that has a formal definition in a few grammar formalisms. In contrast, "part of speech" is the intuitive class of a word. I suppose the equivalent term for syntactic category in HPSG would be Typed Feature Structures or just signs.
In CCG, for example, the syntactic category
S\NP refers to many things: intransitive verb, transitive verb with the object filled-in, a modified verb phrase, etc.
As another example, given the phrases "young Obama" and "Barack Obama", a (trained) person may tag "young" as adjective and "Barack" as noun. However, in CCG, they are both
parts of speech and
syntactic categories have already been given by Tim Osborne. I would like to add a point on whether there is any evidence to label a word a
noun or a
preposition, since this is also occurred in question.
Let's say we already know that table is a noun. Then we can take a sentence where this word occurs and try to replace it with other words. If the sentence is still grammatical then the new word is also a noun. For example,
- The mouse is under the table.
- The mouse is under the bucket.
- The mouse is under the behind.
bucket is also a noun since sentence 2 is grammatical, behind is not a noun since sentence 3 is not grammatical. In this way all words can be assigned to parts of speech.