How can you tell whether prefixed ‘in-’ is the preposition ‘in’ or Indo-European ‘in-’?

Question

Background

The verb īnsum has the prefix in-. Prefixing in/in- to words, changes their meaning to ‘in’, ‘on’ et sim., or ‘un-’, ‘non’ et sim. (ɔ:¹ negation).² However, according to Wiktionary, the pronunciation of in changes in some context; this supported by Lewis.

At first, my understanding was that this pronunciation change – ɔ: from [ɪ] to [ĩː] – only occurs where the prefix is not the preposition in, but the actual prefix in-. Even though sounding the same, they have different roots: When meaning ‘in(side)’, it is merely the preposition prefixed; when meaning ‘not’/‘[negation]’, it has the Indo-European origin [*n̥-].5 As it turns out, the pronunciation change is merely a matter of morpheme interaction.

Non of my grammars, however, even the very detailed grammar by Nils Sjöstrand (Gleerups forlag, Malmö/Lund 1960) has any details on these two differences. The only thing I can find that is mildly relevant, is in Sjöstrand § 6.2, explaining the enlonging [I’m sorry, I am not sure what the correct linguistic term is] and nasalisation of vowels in front of ns and nf.

Question

This means that pronunciation alone is not a clue as to whether or not one is dealing with in prefixed, or the prefix in-, as they both will have their pronunciations changed based on which morphemes follow. How can you know whether you are dealing with the preposition or the prefix, and thus the correct meaning of the word?

Somewhat Related

• Understanding vowel quantity in fieri
• vowel length in “pro” before “f”

Notes

¹ This is not a smiley; it is the symbol meaning ‘that is’, ‘may be read as’ and similar.

² Egil Kraggerud & Bjørg Tosterud: Latinsk ordbok, Cappelen, Oslo 1998: ‘in i sammensetning’.

I have added the negation tag, but am unsure as to its relevance. Advice on this is appreciated. It could very well be that the tag details should be updated to include questions such as this one.

Answer 1

The negative prefix in- typically attaches to an adjective, while the prepositional prefix in- typically attaches to a verb. The main complication to this distribution is the existence of adjectives derived from verbs (or at least, participle forms that look very similar to adjectives) and nouns derived from adjectives or from verbs. But if you're looking at a finite verb form such as "īnsum", you can be fairly confident that the prefix in- does not express negation.

For either prefix, the vowel is long before ns or nf, and short otherwise. There is no difference in pronunciation between the two prefixes spelled in-. They also share a number of alternative spellings representing assimilated pronunciations, such as im- ir- il-, which likewise are pronounced the same for both prefixes regardless of the meaning. (Many words can take either an assimilated or unassimilated spelling, as discussed at -NL- and -LL- in Classical Latin ; I'm not sure whether the rates of assimilated spellings differ between the two prefixes.) The negative prefix has a rare form ig- that I don't think occurs as a form of the prepositional prefix.

Finite verbs are very rarely negated with in-

"Prefix Derivation in Latin", by Dana DINU, says that the negative prefix in- "does not apply to verbs".

This is almost always true. (There are slightly more exceptions than I would have guessed). So far, I've found only the following possible exceptions, some or all of which may actually be back-formed from related adjectives (rather than derived from affixing in- directly to the verb):

• ignōscō, "pardon, forgive", is explained in some sources as coming from the negative prefix and (g)nōscō. For example, Lewis and Short says "lit., not to wish to know, not to search into; hence, with esp. reference to a fault or crime, to pardon". I have asked a separate question about ignōscō to try to determine the explanation for this unusual case.

• improbō, "disapprove, condemn". I'm guessing it is a backformation from improbātus < in- + probātus.

• indecet, "be indecent, unseemly", mentioned by Knapp (1917) when discussing ignosco. Looks like a likely backformation from indecēns. This verb seems to be extremely rare.

• inobaudiō/inoboediō, "disobey". I'm guessing it is a backformation from inobaudiens/oediens. Seems rather rare.

  Example: Tertullian's Adversus Marcionem:

  Cum ergo mandato diceret populo, Ecce ego mitto angelum meum ante faciem tuam, qui te custodiat in via et introducat in terram quam paravi tibi, intende illi et exaudi eum, ne inobaudieris eum; non enim celabit te, quoniam nomen meum super illum est

Participle forms can often be negated with in-, but possibly only when they are actually participial adjectives

Here is a quote I found that indicates that in Classical Latin, non was used to negate verbal participles, while in- appears as a negative prefix for participial adjectives. (If you don't understand what those terms mean, it might help to look at my question about how these two categories of participial forms were distinguished in Latin.)

in Classical Latin the synchronic negative of e.g. patiens ‘enduring’ is non patiens, whereas participles with a negative prefix are synchronically non-participial adjectives, e.g. impatiens ‘impatient’.

(p. 30, "Negated Participles in Rgvedic Sanskrit and Proto-Indo-European," by John J. Lowe, Indo-Iranian Journal 54 (2011) 19-38)

If this rule is true, it would make the distribution of non and in- negation in Latin closely analogous to the distribution of not and un- negation in English for participial forms.

Since participle forms can also contain the prepositional prefix in-, there are some cases of ambiguity between the two in- prefixes.

Answer 2

I don't think it's possible to distinguish in meaning "in" from PIE *en and in- meaning "not" from PIE *n̥ from pronunciation alone. It's well known that the /i/ in in- lengthens when followed by certain consonsant combinations such as "ns" and "nf", but as far as I know, that is purely phonetically determined and has nothing to do with the ancestral morpheme.

So, basically, there is no a priori way of distinguishing the two cases.