Abstract
Although some characteristics of incorporating verbs and non-incorporating verbs have been proposed in previous studies, little systematic cross-linguistic research has been done on restrictions on the types of verbs that incorporate nouns. Knowledge about possible verb-based restrictions on noun incorporation may, however, provide important insights for theoretical approaches to noun incorporation, in particular regarding the question to what extent incorporation is a lexical or a syntactic process, and whether and how languages may vary in this respect. This paper therefore investigates to what extent languages restrict noun incorporation to particular verbs and what types of restrictions appear to be relevant cross-linguistically. The study consists of two parts: an explorative typological survey based on descriptive sources of 50 incorporating languages, and a more detailed investigation of incorporating verbs in corpus data from a sample of eight languages, guided by a questionnaire. The results demonstrate that noun incorporation is indeed restricted in terms of which verbs allow this construction within and across languages. The likelihood that a verb can incorporate is partly determined by its degree of morphosyntactic transitivity, but the attested variation across verbs and across languages shows that purely lexical restrictions play an important role as well.
Introduction
In the field of linguistics, Noun Incorporation (NI) refers to a construction where a noun stem and a verbal stem fuse to form a single complex word—for example, expressing "I am wood-chopping" rather than "I am chopping wood." For decades, a central debate has persisted: is NI a syntactic process, a productive rule that should theoretically apply to any verb given the right grammatical conditions, or is it a lexical process, a restricted method of word formation akin to compounding?
While previous scholarship has extensively documented which nouns are likely to incorporate (typically body parts or generic, non-referential objects), the restrictions placed on verbs have remained under-researched. This paper provides the first systematic cross-linguistic investigation into verb-based restrictions on NI, seeking to identify universal patterns and determine the extent to which these restrictions are driven by semantics, transitivity, or idiosyncratic lexical rules.
Methodology
The researchers employed a two-pronged methodology to capture both broad typological trends and fine-grained usage data:
- Explorative Typological Survey: The authors surveyed 50 languages from 50 distinct language families, including Chukchi, Mohawk, Nahuatl, and Western Frisian. By consulting reference grammars and existing literature, they identified 526 unique verb meanings documented as allowing NI.
- Questionnaire-Based Corpus Study: To observe NI in natural speech, the study focused on eight specific languages with available spoken corpora: Baure, Ese Ejja, Guarayu, Iraqw, Kalamang, Movima, Plains Cree, and Yucatec Maya. Experts analyzed these corpora using a questionnaire of 47 specific verb meanings (e.g., EAT, KILL, JUMP, BE DRY) to determine not just the possibility of incorporation, but its frequency in discourse.
To standardize the analysis, the verbs were evaluated against the "Object dem/del hierarchy" and "Transitivity prominence" scores—metrics that rank how "transitive" a verb's meaning is perceived to be across global languages.
The Transitivity and Semantic Profile of Incorporating Verbs
The study reveals that a verb’s likelihood to incorporate is heavily influenced by its position on the transitivity scale. The "sweet spot" for NI involves two primary types of verbs:
- High-Impact Transitives: Verbs like WASH, CUT, and EAT, which involve a significant impact on an object, are the most frequent incorporators.
- Patientive Intransitives: Verbs where the subject is affected by a state or change of state (e.g., BE DRY or FEEL PAIN) also show high rates of incorporation.
Conversely, several semantic barriers prevent NI. Animacy is a major factor; verbs that typically take human or animate objects (such as MEET, HELP, or TELL) almost never allow NI. Incorporation is generally reserved for inanimate, generic, or non-individuated entities. Furthermore, the distinction between Manner and Result is critical. "Manner" verbs (like JUMP or RUN) are poor candidates for NI, whereas "Result" or "Impact" verbs (like BREAK or HIT) are significantly more likely to participate.
Lexical Idiosyncrasy and Institutionalization
Beyond broad semantic rules, the study highlights the role of cultural and lexical factors. NI is frequently used to describe institutionalized activities—habitual or culturally recognized tasks. For instance, a language is more likely to have a dedicated form for "wood-chopping" (a routine activity) than for a random or unique event like "cloud-watching."
Crucially, the data supports the idea that NI is not purely syntactic. The researchers found evidence of lexical idiosyncrasy, where specific words have their own rules encoded in the lexicon. In Iraqw, for example, two different verbs meaning "cut" allow NI, but a third verb with a similar meaning does not. This suggests that even when a verb meets all the semantic criteria for incorporation, its ability to do so may be restricted on a word-by-word basis.
Key Results and Statistical Findings
The study's quantitative analysis confirms several cross-linguistic trends:
- Top Incorporators: Across the 50-language sample, the verbs WASH, CUT, and EAT emerged as the most frequent and universal incorporators.
- Statistical Correlation: The researchers found a moderate, statistically significant positive correlation (ρ = 0.56) between a verb's "transitivity prominence" and its likelihood to allow NI.
- Implicational Hierarchy: The findings suggest a hierarchy of "incorporability." If a language allows NI for low-transitivity verbs (like JUMP), it almost certainly allows it for high-transitivity verbs (like WASH), but the reverse is not necessarily true.
Conclusion and Future Directions
This research demonstrates that noun incorporation is a constrained process influenced by a combination of morphosyntactic transitivity, semantic impact, and lexical tradition. While transitivity provides a strong predictive framework, the presence of idiosyncratic gaps suggests that NI remains, at least in part, a lexical process.
The authors note that future research would benefit from larger, more standardized corpora for indigenous and endangered languages. Because current corpora are relatively small (ranging from ~29,000 to 160,000 words), the absence of a specific incorporated verb form in the data does not definitively prove its impossibility. Additionally, further study is needed to disentangle how the different functional uses of NI—such as "possessor raising" or "information backgrounding"—interact with these verb-based restrictions.
Summary updated