Meaning of star/asterisk in linguistics

Question

In some dictionaries/lexica, I've seen the asterisk in front of old words. What does it mean/stand for?

Example: http://en.wikipedia.org/wiki/Proto-Germanic#Pre-Proto-Germanic

*ǵʰóstis "stranger" > *gʰóstis > *gastiz "guest"

Answer 1

An asterisk is generally used to indicate that a certain form or construction is not found in natural language. To be precise, it means there is insufficient evidence to assume that it could exist or could once have existed in natural language.

When describing proto-languages, this usually means that a certain root or word has been reconstructed: based on phonological rules, we think it must have been somewhat like this—but we cannot be sure, as it is always possible that some unique irregularity would result in a different form, and we have no written sources that contain this form. When a form has never been found in a real source, we say it has not been attested; in other words, it is unattested. In most proto-languages, all forms are hypothetical, so that they should all be preceded by an asterisk.

With modern languages, we usually have plenty of sources to establish whether a certain form is possible. In fact, many linguists make up example sentences when they need to, because many forms and constructions are not at all controversial. So I might use Achilles hated Hector in an article if I needed an example. However, I may also want to use an example that is ungrammatical, i.e. that I believe would never be used naturally in the language under observation. Then I would put an asterisk in front of it:

*Achilles did hated Hector.

Note that it depends on context whether a construction is grammatical: if I were writing about standard English, I'd have to use an asterisk; but, if I were writing about a certain dialect where this construction is actually used by some, the asterisk is out of place.

Note also that "natural language" is a flexible and sometimes unclear concept; if you are researching English poetry of the 16th century, it is much harder to acquire enough evidence than with modern prose. Context or explanation should make it clear what natural language is supposed to be in a certain text.

---

Most people would use a question mark instead of an asterisk with well-attested languages if the form or construction is doubtful. In proto-languages, the question mark is sometimes used to indicate a form that is even more uncertain than the common hypothetical ones—for example, if I am in doubt as to the most probable form of a certain root.

Some people use the percentage sign to mark something that is only grammatical in a certain non-standard dialect.

Answer 2

It's true that an asterisk indicates a reconstructed word or morpheme in historical linguistics.

In theoretical linguistics, as Cerebus suggested, it indicates ungrammaticality. There are a few other such markers: question marks indicate relative unacceptability, and hash marks indicate anything from semantic incoherence (ex. (1c)) to infelicity, a pragmatic failure (ex. (1d)).

1.a. *John slept the baby.

b. ?John ate much rice.

c. #Colorless green ideas sleep furiously

d. #His horse sleeps until dawn. [In response to the question "How are you doing?"]

Answer 3

In the context of historical linguistics, it means the word is just a hypothetical reconstruction, not an actually attested word.

(The asterisk has another meaning which is to mark an ungrammatical utterance)