What is the grammatical "logic" of impersonal constructions like "Me non solum piget stultitiae meae sed etiam pudet" (Cic. De Dom. 29)?

Question

What is the grammatical "logic" of the impersonal construction with psychological verbs like pudet, piget, paenitet, taedet, miseret? (here is a short descriptive characterization of so-called “psych verbs” (in English))? A couple of typical examples follow:

Me non solum piget stultitiae meae sed etiam pudet. (Cic. De Dom. 29)

Sunt homines quos libidinis infamiaeque suae neque pudeat neque taedeat. (Cic. Verr. I, 1, 35)

Why is the experiencer object marked with accusative rather than with dative? (e.g., cf. Italian: A me mi importa solo di questo. Gloss: 'To me (dative!) matters only of this'. Translation: ‘Only this matters to me’). As for the genitive case, I guess that such an unexpected marking could be related to that of other mental verbs like memini. Many years ago I read (unfortunately, now I can’t remember the reference) that psychological verbs in Indo-European were basically denominal (i.e., derived from nouns), which in fact makes sense (e.g., in some languages something like “I fear ghosts” is expressed by means of a locative construction with the psych emotion being expressed with a noun: “the fear towards ghosts is (placed) in me”. Irish is an example of these languages).

Assuming that the Latin psychological verbs at issue are denominal, I was wondering if the genitive case could be explained by claiming that it is actually modifying the noun incorporated in the verb. But what about the accusative case of the experiencer object? In principle, it is a bit surprising since accusative marking of the object would typically require having a semantic/meaningful (i.e., non-expletive) subject in the sentence: e.g., cf. so-called ‘Burzio’s Generalization’. Notice that the Latin examples at issue (unlike examples with dative experiencer like Hoc mihi placet or with accusative experiencer like Hoc me delectat) can be regarded as involving an intriguing violation of this linguistic generalization. In this impersonal construction the verb does not assign a semantic function to its subject but it does assign accusative case to its object.

Another curious well-known fact of these verbs is that the construction is not impersonal if the subject is a pronoun (e.g., Non te haec pudent? Ter. Adelph., 754) or a clause or an infinitive (e.g., Me, mi Pomponi, valde paenitet vivere. Cic. Att. 3,4). As for the latter example, I was wondering why a gerund in genitive case could not instead be used here: e.g., Me paenitet vivendi.

Answer 1

Let us look at the verbs behaving like this and some corresponding nouns:

1. pudere; pudor
2. paenitere; paenitia, paenitudo
3. taedere; taedium
4. pigere; ?
5. miserere; miseria

The list of related nouns is probably not exhaustive, but I could not come up with more. In cases 1–3 all the nouns appear to be derived from the verb rather than the other way around. I found nothing related to pigere. In case 5 both the verb and the noun appear to come from the adjective miser.

Above is a list of five common verbs with the peculiar behaviour you describe, but none of them appears to be derived from a noun which survives in classical Latin. Of course, earlier nouns cannot be ruled out, but that sounds rather unlikely. Therefore I am tempted to dismiss the theory of denominal origins for verbs of this kind unless some proof is produced.

It is worth noting that all these verbs are of the second conjugation, and verbs in that conjugation tend to be semantically different from others. They often describe state rather than action. Some of these verbs are derived from adjectives, like ruber > rubere and albus > albere. Therefore one might want to analyze 4 and 5 above as miser > miserere and piger > pigere. But this doesn't quite work; the meanings are off and the roles of the accusative and the genitive are too different. Therefore I cannot see these kinds of verbs as state verbs derived from adjectives either. Nevertheless, these verbs do describe state rather than action as is common in their conjugation.

I am not convinced there is a grammatical logic that can be justified by the origins. There is a grammatical logic that I use to make sense of these verbs to myself, and I have no evidence for it beyond its making sense to me. I would be glad to hear of other approaches than the one I will describe next.

These verbs are impersonal. Just like with pluere, there is no subject. The verb pudere means "to make ashamed", "to excite/cause shame", or similar. Therefore me pudet is "[it] makes me ashamed". Such wording is unnatural in many languages (not Finnish!), so it is often best translated in a different way. Better English would be "I am ashamed". The cause of this emotion is expressed as a cause, not as a subject. A cause is typically expressed with ablative, not genitive. This can be circumvented by inserting an ablative like causa. This turns me pudet stultitiae [causa] into "[it] makes me ashamed because of stupidity".

(There is also a question about miserere mei.)

Answer 2

Charo's question has reminded me of this old post of mine, where I was asking for an explanation of the "peculiar" grammar of the impersonal me pudet construction. This question can be divided into two subquestions: (i) why is the Experiencer marked with the accusative case and (ii) why is the Stimulus marked with the genitive case? I.e. is there any (logical, diachronic or whatever) explanation for the fact that the impersonal construction Me pudet stultitiae meae was used? Cf. the ungrammatical (not just unattested) construction Stultitia mea me pudet, where the Stimulus/Cause stultitia mea is feminine singular nominative (but see the last paragraph of my question above).

In what follows I will try to show why I think the following interesting work by Ranko Matasović can shed light on both subquestions:

Matasović, Ranko (2013). «Latin paenitet me, miseret me, pudet me and active clause alignment in Proto-Indo-European». Indogermanische Forschungen. Zeitschrift für Indogermanistik und historische Sprachwissenschaft 118: 93-110.

Matasović (2013: 106-108) claims that the argument structure pattern of Latin impersonal psych verbs of the pudet-type (i.e. 'Source causes the sensation X on Experiencer') is a blend/merge of the two following ones:

• Argument structure of causative constructions: 'Causer (Nom) causes the mental state X on the Experiencer (Acc)'
• Argument structure of stative constructions: 'Experiencer (Nom) feels the sensation X (with respect to the Source (Oblique= Abl/Gen/Acc))'.

As noted by Matasović (2013: 93), it is important to recall that verbs with case-frames like pudet , paenitet, etc. belong to the second conjugation, which includes the PIE causatives (e.g. PIE *mon-eye-ti ‘admonishes’ > Lat. monet) and statives (PIE *tak-eh1-ti ‘is silent’ > Lat. tacet). This is the consequence of the regular sound changes (*eh1 > ē, *eye > ē) by which the stative and causative suffixes merged as Latin -ē-. This author argues that the irregular case-frames of the Latin bivalent statives are innovations based on the analogy with the case-frames of causative verbs, which had the Causee argument in the accusative case.

That is to say, according to him, the accusative marking of the Experiencer of impersonal pudet-type verbs is related to the causative nature of these verbs. In me pudet tui, the accusative marking of me can then be due to its expressing the Causee argument of causative verbs (cf. also docere or monere).

In my opinion, to relate the accusative case marking of the Experiencer in pudet-type verbs with their causative nature is quite insightful. However, I disagree with Matasović's (2013) semantic paraphrase of Lat. impersonal psych verbs: 'Source causes the sensation X on Experiencer'. My proposal would rather be the following one: 'It causes the Experiencer (Acc.) to have the sensation X (root) from Y (Gen.)'.

That is to say, Matasović’s (2013: 106-108) paraphrase of Lat. impersonal psych verbs (i.e. 'Source (Gen) causes the sensation X on Experiencer (Acc)') seems to be based on the wrong premise that the Stimulus has structural preeminence over the Experiencer. However, as pointed out by Devine & Stephens (2013: 123), "the stimulus is lower in the event structure than the experiencer (...) That is why a reflexive can be bound by the object argument" (italics mine: Mitomino).

senectutis eum suae paeniteret (Cic. Cato 19) ‘He would not repent of his old age.’

quos libidinis infamiaeque suae neque pudeat neque taedeat (Cic. Verr. 1, 1, 35) ‘Who are neither ashamed nor upset by their own licentious behavior and bad reputation.’

So the causative nature of the Latin impersonal construction Pudet me tui (lit. 'It causes me to be ashamed of you') can be claimed to run parallel to the one of examples like the one below from Russian (lit. 'it causes me to be nauseated'). See also Joonas's interesting remarks on stative causative constructions in Finnish (see the comments in his answer).

Menja tošnit. (Russian) gloss: me.ACC nauseate.3SG (ex. from Pesetsky (1995: fn. 119, p. 311))

Some interesting discussion on examples like the Russian one above can be found in Pesetsky (1995: 111-113) section on so-called “Emotional Weather” (e.g. see his parallelism between It rains and It’s boring in there):

Pesetsky (1995: 111-113): “Emotions like surprise, annoyance, and amusement are indeed like the weather in a number of respects. They are ‘global’ (ambient), affecting one’s perceptions as well as actions. They are transitory. They are somewhat unpredictable in their onset, intensity, and duration. Most important for our purposes, the proximate cause of both weather and emotions can be viewed as a force of nature (...). One possibility might be a requirement that the lexical item used to express natural forces in meteorological and psychic weather sentences must somehow be minimally expressive. An appropriate notion of minimal expressivity might explain why the form used for meteorological natural forces is (in languages that use the strategies seen in English, French, and Italian) generally identical to that used for expletives: third person, neuter if available (otherwise masculine), null pro(noun) if available (otherwise an unstressed pronoun).” (bold mine: Mitomino).

Pesetsky, David (1995). Zero Syntax. Experiencers and Cascades. Cambridge, Mass.: MIT Press.

---

Now let me address the next question: why is the Stimulus marked with the genitive case?

As noted in my question above, some authors have claimed that the root of Lat. psychological verbs has a nominal nature in PIE. E.g. see Bouchard’s (1992: 33; fn. 10) insightful proposal (attributed by him to Roger Higgins, p.c.) that psychological verbs are denominal in PIE. [See Bouchard, Denis (1992). «Psych Constructions and Linking to Conceptual Structures». In Hirschbühler, Paul; Koerner, Konrad (ed.). Romance Languages and Modern Linguistic Theory. 25-44. Amsterdam/Philadelphia: John Benjamins].

Recall that nomen in Latin holds for both nomen substantivum and nomen adjectivum (unsurprisingly, we also use declinations for adjectives: they are nomina). I am indeed aware that Bouchard's/Higgins's remark above can be taken at most as an interesting speculation but note that Matasović himself relates the original roots of pudet-type verbs to nomina:

Matasović (2013: 98-99): “Lat. miseret ‘feel sorry for’ is derived from miser ‘poor, unfortunate’ [...]. Lat. paenitet ‘cause to regret’ is derived from paene ‘almost, practically’. It appears to contain a root *payn-, the original meaning of which should have been ‘lack’ [...]. Lat. piget ‘affect with revulsion, irk’ is from the same root as piger ‘torpid, inactive’ [...]. Lat. pudet ‘be ashamed of’ is derived from the same stem as pudor ‘shame, decency’ [...]. Lat. taedet ‘be tired of’ is probably connected to taeter ‘foul, horrible’”.

So, to the extent that there was a nomen (substantivum or adjectivum) incorporated in the psychological verb, the genitive case marking of the Stimulus can be claimed to be related/traced back to this fact (genitive is the typical case with which complements of nomina are marked. NB: some linguists have considered the genitive case marking of verbal arguments as an "anomaly" of the grammatical system: genitive is an adnominal case, not an adverbal one). As I said to Joonas in the comments section of his answer, in my opinion, Se non è vero, è ben trovato.