Derived Nominalizations

What is a deverbal nominalization?

Jane Grimshaw’s (1990) seminal study of deverbal nominalizations showed that they were systematically ambiguous between two readings. She called the two readings “result nominals” and “complex event nominals” and provided the following example of examination to illustrate the difference (adapted from Grimshaw 1990:49):

• (424) a. Lydia’s examination was on the table. (result nominal)
• b. Lydia’s examination of the patients took a long time. (complex event nominal)

Grimshaw defined the two as follows: “Result nominals name the output of a process or an element associated with the process; process nominals name a process or an event” (1990:49). On the same page, she renames “process nominals” to “complex event nominals.” Proper naming is a contentious issue with deverbal nouns. Bernard Comrie’s (1976b) term “action nominal” is problematic, since not all verbs from which nouns can be derived refer to actions, and it does not reference the systematic ambiguity of these kinds of nouns. Hagit Borer (2013) renamed Grimshaw’s categories into “R(eferential)-Nominal” and “A(rgument)S(tructure)-Nominal.” There are a few advantages to this description. First, not all “result nominals” refer to results, but they are referential in that they refer to an entity, whether physical or abstract. Second, the term AS-nominal references the syntactic arguments normally present with these kinds of nouns rather than the kind of eventuality. Again, not all of these nominals refer to events (e.g. appreciation from the state appreciate), but when they refer to the same eventuality as the underlying verb, they normally have the same argument structure. Rochelle Lieber (2016), however, challenged the idea that AS-nominals are always found with their arguments, so she went back to the term “Event-nominal” (which we can modify to include states with “E(ventuality)-Nominal”).1 Finally, Lieber noted that there are not two different words pronounced as examination in examples like (1a) and (1b), but there are two different readings of the same noun. Thus, she speaks of R-readings and E-readings of nominals instead of R-nominals and E-nominals. This also is a helpful corrective.

In referring to the phenomenon under discussion, we will follow Lieber (2016) and use the terms R(eferential)-reading and E(ventuality)-reading to describe the interpretations of these nouns. The above “digression” on naming shows how the literature has described these nominalizations. The R-reading of deverbal nouns is associated with results or things that are related to the eventuality named by the underlying verb. In (1a), the predicate was on the table forces the R-reading of examination, and in this case, it refers to a physical object, specifically an object on which the examination is

4.0 International License. https://doi.org/10.1515/9783111713632-057 printed. The R-reading lacks the argument structure of the underlying verb examine, which normally takes a direct object. The initial possessive Lydia’s must, therefore, be interpreted as someone who possesses the examination and not someone who engages in the event of examining.

When we come to the E-reading, the argument of the verb examine (normally a direct object) surfaces as a genitive phrase of the patients; this is why Borer termed this reading an Argument Structure-nominal – the nominal shares the same argument structure as the underlying verb. There are other verb-like features in this reading, namely protraction over time, (i.e. it took a long time), the possibility of adding in the original subject of the verb (e.g. Lydia’s examination of … where Lydia refers to the original subject), and possible adverbial modification of the eventuality (e.g. Lydia’s (frequent) examination of …). The fact that these verbal features are present suggests that the underlying verb is playing a significant role in the interpretation of deverbal nouns, at least in the E-readings. Given the principle of compositionality, this is expected .

Although less obvious, R-readings are also largely predictable from their underlying verbs. Melloni (2011) investigates R-readings of nominalizations from different verb classes, and she notes that R-readings do not necessarily refer to true results. Some verb classes, like verbs of creation, do in fact make nominalizations with R-readings that refer to results. For example, build refers to the process of constructing something, and the related nominalization building may refer to this process with the E-reading (Building the house took 6 months) or to the result of that process with the R-reading (The building is 10 feet tall). Verbs in other classes create different nominalizations. Many of these, however, can fit under the general category of an entity in state reading. In other words, the R-reading of the nominalization refers to an entity in a state that is related to the eventuality described by the verb. Thus, building refers to an entity in the state of having been built. Likewise, the R-reading of assembly refers to the entity in the state of having been assembled, destruction refers to the entity in the state of having been destroyed, preference refers to the entity in the state of being preferred, etc. Some of these “entities” are more abstract than others, but they all refer to “things” that are related to the underlying verb.

The English nouns we have used to illustrate this phenomenon might have similar readings when compared to their underlying verbs, but they all have different suffixes. The - ing (building), - y (assembly), - tion (destruction), and - ence (preference) suffixes each have their own unique characteristics. In Lieber’s (2016) comprehensive study of nominalizations in English, she notes that the kind of R-readings found with each suffix varies. Not all suffixes allow for the E-reading, and the R-readings available are usually limited, since the suffixes only combine with certain classes of verbs. Even a suffix with a broad distribution like - ing lacks 9 of the 15 available readings in all the nominals it combines with. Other suffixes like - al have only three readings (examples from Lieber 2016:63): an E-reading (… Achebe’s portrayal of duality), a result R-reading (e.g. … seven years after Jackson’s acquittal), and an inanimate patient reading (… get the rental). Comprehensive coverage of both the form and meaning of nominalizations in English can be found in Bauer et al. (2013) and Lieber (2016).

Syntactic characteristics of deverbal nominalizers in Greek

When we turn to Greek, we see that some nominalizers (suffixes that turn something into a noun; see Morph A) are polysemous between an E-reading and R-reading, as expected. (A polysemous word has two distinct interpretations, only one of which is normally correct in the context.) The examples below demonstrate this ambiguity (1 Pet 4:13 is first given in 40.2.3.2):

• (425) E-reading of ἀποκάλυψις
• ἵνα καὶ ἐν τῇ ἀποκαλύψει
• So that you may rejoice also in the revealing of his glory
• (426) R-reading of ἀποκάλυψις
• ἀποκάλυψις Ἰησοῦ Χριστοῦ ἥν ἔδωκεν αὐτῷ ὁ θεός (Rev 1:1)
• The revelation of Jesus Christ which God gave to him

In (425), ἀποκάλυψις refers to the event of Jesus’s glory being revealed, so this is the E-reading of the derived nominal. Revelation 1:1, on the other hand, illustrates the R-reading of ἀποκάλυψις, since an event cannot be ‘given.’ Rather, ἀποκάλυψις refers to an (abstract) entity, namely the thing revealed. Although this shows that nominalizations in Greek exhibit the same kind of ambiguity as nominalizations in English, we would also want to know what kinds of syntactic characteristics, if any, are associated with each reading and whether different suffixes are associated with particular readings.

One syntactic difference between (425) and (426) is that the former noun is found with the article, and the latter is not. The noun in the genitive phrase is also definite in the E-reading (τῆς δόξης αὐτοῦ), while the R-reading is not (Ἰησοῦ Χριστοῦ). This is true in Rev 1:1 even though the revelation given is a single unique revelation and is, therefore, best translated into English with the definite article ‘the’ rather than an indefinite article ‘a’. In Grimshaw’s (1990) original study, she claimed that E-readings are only available with definite determiners (like the) and are not found with indefinite determiners (like a). This accords with the data in (425) and (426), but of course, we cannot conclude from these data that E-readings are always found with the article, and given all the mismatches between the use of the definite article in Greek and English (see Semantics B), we would need to conduct our own analysis of E-readings in Greek to determine to what extent this is true. However, Ernest De Witt Burton’s description of the subjective genitive with πίστις in his commentary on Galatians conforms to this rule (1921:482):

Those in which πίστις is accompanied by a subjective genitive or equivalent phrase indicating by whom the faith is exercised. The article is in this case almost invariably present. The object of the faith is usually indicated, more or less definitely, by the context, but occasionally directly expressed …

We return to the genitive with πίστις in Exegesis D. In Grimshaw’s original study (1990:54), she posited that E-readings can only occur with definite determiners, which would agree with Burton’s description. Lieber’s data (2016:51), however, suggests that Grimshaw was incorrect, and, indeed, there do seem to be E-readings with nominalizations in the Greek of the NT as well (to be clear, Grimshaw’s claim would have only applied to English):

• (427) E-reading without the article
• εὑρεθῇ εἰς ἔπαινον καὶ δόξαν καὶ τιμὴν ἐν ἀποκαλύψει Ἰησοῦ Χριστοῦ (1 Pet 1:7)
• … may be found to result in praise, glory, and honor in the revealing of Jesus Christ

Even this counterexample, however, is not the best. Proper names sometimes behave differently syntactically, and the only two clear examples we have of ἀποκάλυψις without the article are with the name ‘Jesus Christ (1 Pet 1:13 being the only other example). All other examples of E-readings with ἀποκάλυψις have the article (Rom 8:19; 1 Cor 1:7; 2 Thess 1:7; 1 Pet 4:13), and most R-readings lack the article (Luke 2:32; Rom 16:25; 1 Cor 14:6, 26; 2 Cor 12:1; Gal 1:12; 2:2; Eph 1:17; 3:3; Rev 1:1).2 This is certainly not to suggest that R-readings never have the article – 2 Cor 12:7 is an R-reading that has the article. However, these data do reveal (pun intended) a robust tendency: E-readings in Greek nominalizations normally have the article, and R-readings normally lack the article. A complete description would cover in what contexts we see exceptions to this (such as a proper name as the argument) and whether other nominalizations also conform to this generalization.

The other major syntactic characteristic of E-readings is that the argument(s) of the underlying verb is syntactically present. Again, all E-readings conform to this. The thing revealed at a certain time is explicitly given in the syntax. With ἀποκάλυψις, the thing revealed, normally the object of active ἀποκαλύπτω or the subject of passive ἀπεκαλύφθην, is found in the genitive. No E-reading has what would be the subject of the active verb. There are theoretical reasons for this that need not concern us here, but the principle noted in the literature that E-readings contain arguments generally holds for Greek nominalizations as well. However, this does not mean that every genitive following ἀποκάλυψις must be interpreted as an argument of the underlying verb ἀποκαλύπτω, as shown by the following pair:

(428) E-reading with genitive as argument

• ἀπεκδεχομένους τὴν ἀποκάλυψιν τοῦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ (1 Cor 1:7)
• Waiting for the revealing of our Lord Jesus Christ
• (429) R-reading with genitive as non-argument
• οὐδὲ γὰρ ἐγὼ παρὰ ἀνθρώπου παρέλαβον αὐτὸ οὔτε ἐδιδάχθην, ἀλλὰ δι’ ἀποκαλύψεως Ἰησοῦ Χριστοῦ (Gal 1:12)
• For I neither received it nor was taught it from a person, but through a revelation of Jesus Christ

2 I omit Rom 2:5 which has textual issues. As the text of the NA28 stands, it would be an E-reading without the article.

In (428), the thing the Corinthians are waiting for is the event of Jesus being revealed. It is normal to wait for a particular event to occur, so the E-reading is most salient here, and this is further confirmed by the presence of the article in τὴν ἀποκάλυψιν. The following genitive phrase, then, refers to a participant in the event of revealing, so it would be an argument of the underlying verb ἀποκαλύπτω (and it also is articular). In contrast, the ἀποκάλυψις referred to in (429) is something that would teach Paul. This cannot be an event, since we do not learn things from events. However, we can learn from the “thing” an ἀποκάλυψις might refer to, which would include content of some kind (see 14.2.3.2), so Gal 1:12 would have an R-reading. The genitive, then, would give what the content is about and not the participant in an event, so it would not be an argument of the underlying verb ἀποκαλύπτω.

The presence or absence of arguments is often difficult to determine. Arguments can be deleted phonologically but may still be present syntactically and semantically. It is also not the case, as we have seen, that any genitive phrase can be assumed to be an argument, and arguments may also be represented in different ways than genitive case (again, see Exegesis D). Despite these caveats, this characteristic of E-readings is, like the presence of the article, quite robust: the argument(s) of the verb is normally present in the E-reading of a deverbal nominalization.

Greek nominalizers and their meanings

Ron Moe, who has worked on a NT Greek dictionary for many years as of the time of writing (a project which I have slowly been helping with), provides a list of all the nominalizers found in NT Greek (that is, all suffixes that create nouns). The following are all suffixes in our database that can derive nouns from verbs:3 -α* (διασπορά ‘dispersion’ from διασπείρω ‘disperse’), -αλο (διδάσκ αλο ς ‘teacher’ from διδάσκω ‘teach’), -ες* (εἶδο ς ‘appearance’ from εἶδον ‘see’), -η (οἰκοδεσπότης), -θμο (κλαυθμός ‘weeping’ from κλαίω ‘weep’), -κη (ἀποθήκη ‘barn’ from ἀποτίθημι ‘put away’), -ματ* (ἀγνόημα ‘mistake’ from ἀγνοέω ‘not know’), -μο* (ἁγιασμός ‘consecration’ from ἁγιάζω ‘consecrate’), -μον (ἡγεμών ‘leader’ from ἄγω ‘lead’), -ο* (δάκρυον ‘tear’ from δακρύω ‘cry’), -ον (εἰκών ‘image’ from ἔοικα ‘resemble’), -οντ (ἄρχων ‘ruler’ from ἄρχω ‘rule’), -ς (κλέος ‘fame’ from κλέω ‘make famous’), -σι* (ἀποκάλυψις ‘revelation’ from ἀποκαλύπτω ‘reveal’), -σμο (δεσμός ‘shackles’ from δέω ‘bind’), -τ* (γενετή ‘birth’ from γεννάω ‘give birth’), -τηρ (νιπτήρ ‘washbasin’ from νίπτω ‘wash’), -το (διάλεκτος ‘language’ from διαλέγομαι ‘talk’), -τορ (ῥήτωρ ‘orator’ from ἐρῶ ‘will say’), -τρο* (λύτρον ‘ransom’ from λύω ‘loose’), and -ων (ἀγών ‘race’ from ἄγω ‘lead’).

All suffixes marked with *, namely -α, -ματ, -ο, -σι, and -τ, are found with 20 or more distinct lexemes showing they are particularly productive. Some suffixes could arguably have been included that did not, strictly speaking, form a noun from a verb. For example, the suffix -μη is found in the word ἀφορμή ‘opportunity’, which is a combination of ἀπό + √ὁρ + -μη. This noun is related to the verb ἀφοράω ‘watch’ from ἀπό + √ὁρ + -α. The reason why this was not included was both the noun and verb come from the same stem, namely ἀπορ-. Another excluded case was στάμνος ‘jar’ from √στα + -μνο even though √στα seems to be a root that refers to an eventuality (‘stand’). Roots are not themselves verbs, so any word formed from a root directly was not included.

The suffixes that are less productive generally have less readings than suffixes that are more productive. For example, -τηρ is limited to instruments (νιπτήρ ‘washbasin’) or agents (σωτήρ ‘savior’) related to the underlying verb, and these nominalizations appear to lack E-readings. In contrast, a suffix like -σι combines with a host of different stems that have various R-readings (in addition to E-readings), such as result (κτίσις ‘creation’), quality (φύσις ‘nature’), content noun (ἀποκάλυψις ‘revelation’), patient (βρῶσις ‘thing eaten’), and others. Again, if the specific R-reading is determined from the underlying verb, this makes sense. The broader range of verb classes that a suffix might combine with, the broader range of R-readings we expect the suffix to exhibit.

Just as we left many unanswered questions in the syntax section, so there is much more that could be said about the semantics of deverbal nominalizations in Greek. A full analysis would include a description of all the different verb classes that combine with the various nominalizers and what kinds of E and R-readings result. We would also want more comprehensive coverage of the morpho-syntactic processes found for each affix (such as suffixation vs. conversion), and a

3 It should be noted that not all free forms of the suffixes appear the same as what we find in words, e.g. -ες looks like -ος in εἶδος. This is due to morpho-phonological rules particular to the form of the suffix and the stem to which the suffix attaches (for some of these, see Phonology B).

detailed description of how the meanings are reflected in the syntax would also be called for. These lacunae notwithstanding, we have provided enough of a description to analyze the deverbal nominal πίστις in the so-called “pistis Christou” debate.