-
Notifications
You must be signed in to change notification settings - Fork 4
MatrixDoc_Nominalized Clauses
This document explains how to fill out the Nominalized Clauses page of the Grammar Matrix Customization questionnaire. This page allows the user to define a nominalization strategy which can then be applied to verbal LRTs on the Morphology page. Any verb which takes an affix marked by a specific nominalization strategy will have the properties specified for that strategy.
For each defined nominalization strategy, the user is required to enter a name. Once a least one nominalization strategy has been defined, a nominalization feature will become available on the Morphology page for all verbal position classes. The values of this feature are the names of all defined nominalization strategies. Each nominalization LRT then uses one of the defined nominalization strategies. It is not possible for a single LRT to use more than one strategy.
A strategy's properties are primarily determined by its nominalization type. Four of the types are based on the typology presented in Koptjevskaja-Tamm (1993) which organizes ANCs into four categories based on how the arguments referring to the agent and the patient of the underlying verb are marked. These four types are sentential, poss-acc, erg-poss, and nominal. Descriptions of these four types as well as two additional categories (alternative-sentential and all-comps) which serve as slight extensions are given below. In the following descriptions A refers to the more agent-like argument and P refers to the more patient-like argument. Each of the below sections provides information on how to specify constraints on the arguments of the nominalized verb which are available on the Morphology page (subject, object, second object). Information regarding how to specify marking on arguments not available on the Morphology page (syntactic possessor arguments) is given in its own section.
The sentential type describes nominalized clauses where all dependents are marked in the same way as they would be with the non-nominalized verb. For example, Godoberi [god] finite clauses require transitive subjects to take ergative case while intransitive subjects and direct objects take absolutive case. As the below example shows, even in nominalized clauses, this same pattern of case-marking is observed with aHmadi 'Ahmad' (the A) taking ergative case and rec'i 'bread' (the P) taking absolutive case (Koptjevskaja-Tamm, 2013).
| aHmadi-di | maHamadi-łi | rec'i | ik̄-ir |
| Ahmad-ERG | Mahamad-DAT | BREAD.ABS | give-NMLZ |
| 'Ahmad's giving bread to Mahamad.' [god] (Kazenin, 1996: 160) |
LRTs marked with a sentential nominalization strategy:
On the Morphology page, the A for this strategy is the subject and the P is the object. All constraints on the subject and object will be identical to those specified on the arguments of the underlying verb.
In this nominalization type, The P is marked in the same as the original verb, but the A is marked differently. This can be either due to being marked with a different case or a different adposition compared to the underlying verb. Crucially, the marking strategy on the A is not covered by any of the possessive strategies in the language. In the below English [eng] example, the P the evidence is marked in the same manner as a typical verbal complement in English (with accusative case) while the A him takes accusative case. This contrasts with main clauses where the subject of a verb takes nominative case.
| he | was | shocked | by | him | destroy-ing | the | evidence |
| he[NOM] | was | shocked | by | him[ACC] | destroy-NMLZ | the | evidence |
| 'He was shocked by him destroying the evidence.' [eng] |
LRTs marked with a alternative-sentential nominalization strategy:
On the Morphology page, the A for this strategy is the subject and the P is the object. All constraints on the object will be identical to those on the object of the underlying verb. Additional feature constraints (form, case) can be specified for the subject.
In the all-comps nominalization type, the nominalized verb takes two complements corresponding to the A and the P, as well as a determiner, if present in the language. It is possible to select whether the first complement of the nominalized verb is the A or the P argument. The question concerning this choice becomes visible whenever all-comps is selected as the nominalization type of a strategy. In the below English [eng] example, the first complement of the secretary is the P argument and the second complement by the chairman is the A argument.
| he | was | shocked | by | the | dismiss-al | of | the | secretary | by | the | chairman |
| he[NOM] | was | shocked | by | the | dismiss-NMLZ | of | the | secretary | by | the | chairman |
| 'He was shocked by the dismissal of the secretary by the chairman.' [eng] |
LRTs marked with an all-comps nominalization strategy:
For transitive all-comps strategies, the user decides whether the A or P is mapped to the first or second object. For intransitive all-comps strategies, the object is always the A. Additional feature constraints (form, case) can be specified on the object and second object. Nominalized verbs resulting from this strategy do not have subjects and no feature constraints should be placed on the subject.
In the poss-acc type, the A of intransitive and transitive verbs is marked in the same way as a possessor in a specific adnominal possessive strategy. The P is marked as it would be for the non-nominalized verb. As the below example from Meadow Mari [mme] shows, the nominalized verb voz-öm 'write' takes a possessive suffix and the A möj 'I' is marked with genitive case, which is also used to mark possessors in non-derived noun phrases (Koptjevskaja-Tamm, 2013). Meanwhile, the P pis’ma `letter' takes accusative case which is the same case used to mark direct objects in finite clauses (Koptjevskaja-Tamm, 2013).
| möj-ön | pis’ma-m | voz-öm-em |
| I-GEN | letter-ACC | wrote-NMLZ-1SG.POSS |
| 'My writing of the letter.' [mme] (Koptjevskaja-Tamm, 2013) |
LRTs marked with an poss-acc nominalization strategy:
The P for this strategy is the object and the A is a syntactic possessor. All constraints on the object will be identical to those on the object of the underlying verb.
In the erg-poss type, the A of intransitive verbs and the P of transitive verbs is marked in the same way as a possessor in a specific adnominal possessive strategy, while the A of transitive verbs uses an oblique marking pattern. In the below Russian [rus] example, the P sonat 'sonata’ takes genitive case as would the possessor in a non-derived noun phrase (Koptjevskaja-Tamm, 2013). In contrast, the A pianist 'pianist’ takes instrumental case which can also be used to mark agents in passive sentences (Koptjevskaja-Tamm, 2013).
| ispoln-enij-e | sonat-y | pianist-om |
| perform-NMLZ-NOM | sonata-GEN | pianist-INSTR |
| 'The performance of the sonata by the pianist.' [rus] (Koptjevskaja-Tamm, 2013) |
LRTs marked with an erg-poss nominalization strategy:
The A for this strategy is the object and the P is a syntactic possessor. Additional feature constraints (case, form) can be specified for the object on the Morphology page.
Finally, in the nominal type, the A of intransitive and transitive verbs is marked in the same way as a possessor in a specific adnominal possessive strategy, and the P is marked in a manner typical of nominal arguments. This either means that both the A and P take possessive marking, or that the A takes possessive marking while the P takes an oblique form of nominal marking (Koptjevskaja-Tamm, 1993). The below English [eng] example serves as an example. The A enemy's is marked as a syntactic possessor while the P of the city could arguably be analyzed as either a syntactic possessor or an oblique form of marking by the nominal preposition of.
| the | enemy 's | destroy-tion | of | the | city |
| the | enemy POSS | destroy-NMLZ | of | the | city |
| 'The enemy’s destruction of the city.' [eng] |
LRTs marked with an nominal nominalization strategy:
The A for this strategy is the syntactic possessor and the P is the object. Additional feature constraints (form, case) can be specified for the object on the Morphology page. Regardless of whether both arguments are marked as possessors, only the A argument can use a defined possessive strategy (defined on the Adnominal Possession page). The marking of the P can only be specified by case or adpositional constraints on the Morphology page.
poss-acc, erg-poss, and nominal nominalization strategies all allow one argument to be marked in the same way as a possessor in a possessive noun phrase. This can be specified by defining the possessive strategy on the Adnominal Possession page and then selecting that that particular strategy can be used in nominalized clauses on the Nominalized Clauses page. Which argument (A or P) takes possessive marking depends on the nominalization type of the strategy and is given above for all three of poss-acc, erg-poss, and nominal. Currently, all defined poss-acc/erg-poss/nominal nominalization strategies must use the same possessive strategies. Thus, if a language has both a poss-acc and an erg-poss nominalization strategy, it is not possible to restrict a given possessive strategy to be used by the poss-acc strategy, but not the erg-poss strategy. The ability to select a possessive strategy becomes available at the bottom of the Nominalized Clauses page whenever either poss-acc, erg-poss, or nominal is selected as the nominalization type for a strategy. All possessive strategies used in nominalized clauses must also be possessive strategies that can be used by regular non-derived nouns.
Both the A and P argument of a nominalization strategy are optional by default. However, it is possible to make either or both of these arguments mandatory. Selecting the checkbox at the very top of the Nominalized Clauses page causes the OPT feature to become available on the Morphology page. Any A and P arguments that are found on the Morphology page can then be made mandatory by specifying that the relevant argument should be [OPT -]. Syntactic possessor arguments can be made mandatory by clicking on the corresponding checkbox on the Nominalized Clauses page. This checkbox only becomes visible for poss-acc/erg-poss/nominal strategies.
For each nominalization strategy it is necessary to specify whether the strategy acts on intransitive verbs, transitive verbs, or both. If both intransitive and transitive checkboxes are selected, only an LRT corresponding to transitive verbs should be defined on the Morphology page. This LRT should be allowed to take either intransitive or transitive verbs as input. An intransitive equivalent will be created automatically based on the transitive version. In most cases, the intransitive version is identical to the defined transitive one aside from lacking any constraints specified on the object. The one exception is all-comps LRTs, since intransitive all-comps action nominals still take a complement. Thus, for all-comps lrts, an intransitive LRT is created from the transitive one by copying all features aside from those on the second object. As a result, if the user specifies that a particular all-comps strategy acts on both intransitive and transitive verbs, any feature constraints placed on the object of the transitive LRT will also be shared by the single complement of intransitive nominalized verbs using that strategy.
For nominalization strategies of these two types, it is possible to make nominalization purely syntactic or else to include nominalization in the semantic representation of the nominalized verb. The question relating to this choice becomes visible whenever sentential or alternative-sentential is selected as the nominalization type. Leaving nominalization out of the semantics prohibits nominalized verbs from being modified by adjectives, and from appearing in certain argument positions such as subject of a verb, complement of an adposition, and certain verbal complements.
For the three nominalization strategies that allow for possessor-marked arguments, there are three options regarding the inclusion of possessive semantics in nominalized clauses.
- both: Include both the possessive and verbal semantics (the argument marked by the possessive strategy will be interpretated as both a verbal argument and a possessor)
- verb-only: Include only the verbal semantics (the argument marked by the possessive strategy will be represented as only a verbal argument)
- noun-only: Include only the possessive semantics (the argument marked by the possessive strategy will be represented as just the possessor)
All poss-acc/erg-poss/nominal strategies will share the same semantic option. The question regarding this choice becomes visible at the bottom of the Nominalized Clauses page whenever a poss-acc, erg-poss, or nominal strategy is defined.
For all-comps/poss-acc/erg-poss/nominal strategies it is possible for the nominalized verb to appear with a determiner. For poss-acc/erg-poss/nominal strategies the determiner appears in place of the argument marked by a possessive strategy. Whenever one of these four strategy types is selected, a question becomes visible regarding whether determiners are obligatory, optional, or impossible. This question can be ignored for languages without determiners. For all strategies which mark an argument with a possessive strategy, this question concerning determiners is relative to the syntactic possessor. For example, an obligatory determiner means that the nominalized verb must appear either with a syntactic possessor or with a determiner. An impossible determiner means that the action nominal can appear with a Poss, or with no specifier at all, but it cannot appear with a determiner. If the question is left unanswered then the determiner is treated as optional in languages with determiners.
Users are allowed to select a single alternative word order in nominalized clauses for just poss-acc/erg-poss/nominal strategies. This same word order must be shared by any defined poss-acc/erg-poss/nominal strategies. The option to select an alternative word order becomes available at the bottom of the Nominalized Clauses nominal Clauses page whenever a strategy of one of these three types is defined. sentential/alternative-sentential/all-comps strategies all use the same word order as in finite clauses. For all-comps this means that both complements will use the verb object order seen in finite clauses. For example, if a language has SVO (subject verb object) word order, then both complements of a transitive all-comps nominalized verb will appear to the left of the nominalized verb (nominalized-verb object second object). For poss-acc/erg-poss/nominal strategies, users should choose a word order option based on the semantics of the arguments with S (subject) mapping to an A argument and O (object) mapping to a P argument. V maps to the nominalized verb.
In certain languages, whenever a transitive nominalized verb appears with just a single argument, that argument is marked in the same way as a syntactic possessor (Koptjevskaja-Tamm, 2003). This is in spite of how that argument is marked when both the A and P appear overtly in the clause. For example, in Russian [rus] whenever a P or an A argument appear alone with a nominalized verb they take genitive case (the same way possessors are marked). However, when both the A and P appear with the nominalized verb, only the P takes genitive case while the A takes instrumental case (Comrie, 1976). The single possessor marked argument is ambiguous between an A (agent) and P (patient) interpretation. For poss-acc/erg-poss/nominal strategies an additional checkbox becomes visible requiring the single argument of transitive nominalized verb to always be marked as a syntactic possessor (the single-possessor analysis). This single argument will be semantically ambiguous between an agent-like and patient-like reading.
For each nominalization strategy, it is possible to select whether nominalized verbs can be modified by adjectives, adverbs, both or neither. The neither option is selected if both the adjective and adverb boxes are left unchecked.
For any strategy that has arguments that are neither marked by possessive strategies nor take identical marking to the underlying verb, it is possible to select additional case-marking or adpositional-marking (specifically semantically empty adpositions) constraints on these arguments. Changes in case-marking are specified by selecting a particular case for the argument on the Morphology page. To require an argument to take a semantically empty adposition, the adposition must first be defined on the Lexicon page. This is done by defining a semantically empty adposition which takes no feature constraints. The process of defining such as semantically empty adposition (neither case-marking not informational structure marking) automatically assigns a form value to the semantically empty adposition. This form value can be used on the Morphology page to require the argument of a nominalized verb to take a particular semantically empty adposition. A semantically empty adposition's form value is its orthography followed by the string _sem.
Comrie, Bernard. 1976. The Syntax of Action Nominals: A Cross-Linguistic Study. Lingua 40. 177-201.
Kazenin, Konstantin I. 1996. The Syntactic Structure of Action Nominal Constructions. In Kibrik, A. E. (ed.), Godoberi, 160-161. München: Lincom Europa.
Koptjevskaja-Tamm, Maria. 1993. Nominalizations. London: Routledge.
Koptjevskaja-Tamm, Maria. 2003. Action Nominal Constructions in the Languages of Europe. In Plank, Frans (ed.), Noun Phrase Structure in the Languages of Europe, 723-759. Berlin: Mouton de Gruyter.
Koptjevskaja-Tamm, Maria. 2013. Action Nominal Constructions. In: Dryer, Matthew S. & Haspelmath, Martin (eds.) WALS Online (v2020.3) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7385533 (Available online at http://wals.info/chapter/62, Accessed on 2024-09-09.)
Home | Forum | Discussions | Events