So, in linguistic terminology, are the terms "infelicitous" and "ungrammatical" synonymous?
The answer is no.
The focus on acceptability vs. unacceptability also led to a more nuanced
understanding of what these concepts mean. The most widespread view,
proposed originally in Chomsky (1965), is that (un)acceptability may be
influenced by a variety of factors, of which (un)grammaticality is only one.
Under this view, then, "acceptability" and "grammaticality" are not syno-nyms [emphasis mine], in that a sentence that is well-formed according to principles of the
grammar, for instance, may turn out to be unacceptable due to parsing
difficulties, etc. Both concepts are presumably gradient (i.e. acceptability
and grammaticality are both "a matter of degree," in Chomsky's (1965)
terms), but only acceptability is perceived directly. Grammaticality, like
the other factors that contribute to acceptability, can only be inferred
based on the evidence available.
The Cambridge Handbook of Experimental Syntax 2021, p. 8.
Or is the term "infelicitous" a more general term, encompassing the violation of more than just morpho-syntactic rules?
The other way around. "ungrammatical" is more general. "infelicitous" encompasses the violation of more than just morpho-syntactic rules.
For some there can be no acceptability without grammaticality. In
other words, grammaticality is a prerequisite for acceptability: 'An
acceptable utterance is one that has been, or might be, produced
by a native speaker in some appropriate context and is, or would
be, accepted by other native speakers as belonging to the language
in question' (Lyons, 1968: 137). The part of Lyons's definition
that refers to the utterance 'belonging to the language in question'
appeals to grammaticality. This relationship between acceptabil-
ity and grammaticality recurs in his later work: 'an ungrammatical
utterance is one that a native speaker can not only recognise as
unacceptable, but can also correct' (Lyons, 1977: 380). As shown
in the previous paragraph, corrigibility is the criterion for gram-
maticality, whereas de facto use and unproblematicity are the tests
for establishing acceptability.
To decide on the acceptability of a piece of language we refer
not to rules, but to contexts, trying to contextualize the utterance
in question. Try it for yourself. Can you think of a context where
the following could be said?: Pele (the Brazilian footballer) wore a
green dress. Yes, if he were taking part in the Rio carnival celebra-
tions. But the learner who said this was referring to the context
where Pele was playing football, and by 'dress' meant 'shirt' or
'strip'. So this is an error of acceptability. Or can you think of con-texts where I came to London last summer to study the English would
be acceptable? The first question is whether the speaker was an
anthropologist studying the English people or a linguist. The sec-
ond problem is one of location deixis. It is acceptable if spoken
when the speaker is back in London, on a subsequent visit, or
(strangely) if the speaker is outside London but speaking on the
phone or writing to someone who is now in London. It would be
quite unacceptable if spoken to a listener outside London, and
even more so if the addressee was not from London or Britain.
Real and imagined contexts raise the degree of acceptability of
doubtful sentences, but judgements about the grammaticality of a
sentence have to be made in the face of the sentence in isolation, not in a context. For this reason, to decide whether something
is acceptable, even when it satisfies the grammaticality test, is sel-
dom clear-cut and takes some thought, even imagination. Pro-
fessor Willy Haas (p.c.) told of the time when the ungrammatical
and thitherto unacceptable Quadruplicity drinks Procrastination sud-
denly became acceptable when he saw in a newspaper a photograph
of the 'Big Four' statesmen — Stalin, Churchill, Roosevelt and De
Gaulle — assembled to drink a victory toast at the end of World
Carl James, Errors in Language Learning and Use 2013, pages 67n8.
It is important to bear in mind that the notions of acceptability and
grammaticality are not exactly the same, since factors not directly related to
the grammar itself can also affect acceptability. For example, a native speaker
might consider a sentence unacceptable because it contains a word that
the speaker finds obscene or offensive, or because the sentence is so long
they forget how it began, or because the sentence is ambiguous in a way
that might be confusing. In such cases, the speaker’s judgments about
acceptability may not be a fully reliable indicator of grammaticality. Thus,
native speakers’ intuitions provide only indirect evidence about whether a
sentence is grammatical. Since there is no more direct way of finding out
whether a sentence is grammatical, linguists generally rely on speakers’
intuitions about acceptability, despite these limitations.
Victoria Fromkin, An Introduction to Linguistic Theory (2000), page 19.
But even linguists mix up these two terms!
However, in the linguistics literature as well as the second language literature,
these two terms are often used without a careful consideration of their differences.
This is the case even for those who, as Myers (2009a) notes, "should know better" (p.
412). For instance, Schütze succumbs to this temptation in his classic 1996 book on
this topic. "Perhaps more accurate terms for grammaticality judgments would be
grammaticality sensations and linguistic reactions. Nonetheless, for the sake of
familiarity I shall continue using traditional terminology, on the understanding that
it must not be taken literally" (p. 52). Nearly 20 years after Schütze's important book
on the topic, Schütze and Sprouse (2013) acknowledge the inappropriateness of the
term grammaticality judgment:
Speakers' reactions to sentences have traditionally been referred to as grammaticality
judgments, but this term is misleading. Since a grammar is a mental construct not accessible
to conscious awareness, speakers cannot have any impressions about the status of a sentence
with respect to that grammar; rather, in Chomsky's (1965) terms, one should say their
reactions concern acceptability, that is, the extent to which the sentence sounds "good" or
"bad" to them. Acceptability judgments (as we refer to them henceforth) involve explicitly
asking speakers to "judge" (i.e. report their spontaneous reaction concerning) whether a
particular string of words is a possible utterance of their language, with an intended
interpretation either implied or explicitly stated.
Patti Spinner, Using Judgments in Second Language Acquisition Research 2019. Anyone know the page number?