Language contact between Italian and English: a case study on nouns ending in the suffix -ing

1 Introduction

The project described here investigates how English deverbal nominalizations with the suffix -ing have been imported into Italian. In current written Italian there are many such forms, and for some of them a corresponding base verb has also been attested in Italian; (1). All examples are taken from the itTenTen16 corpus (Jakubíček et al. 2013) unless indicated otherwise; a list of the forms under study is given in the Appendix; see Table 1.

The focus of this article is on the semantic characteristics of these borrowed deverbal nouns in Italian, and in particular on the question of whether they have been borrowed not as simple sign-concept pairings but with argument and event structure. Previous research on deverbal nominalization has assumed that argument structure and event structure in a deverbal noun must be licensed by some overt affix (Borer 2014; Grimshaw 1990; Smith 1972). This implies that a loan noun – in case it is attested with argument and event structure – is not borrowed holistically but rather retains its internal morphological structure and, thus, that morphological borrowing of the affix must have occurred. According to Gardani (2021), morphological borrowing of an affix from a source language (SL) to a recipient language (RL) should be said to have occurred only when this affix shows an ability to combine with native elements of the RL; Gardani (2021: 134) considers combinations with non-native bases not to be morphological borrowing. In Italian, the distribution of the suffix -ing is restricted to English bases, i.e., to bases from the SL; hence, no morphological borrowing has occurred. English nominalizations ending in -ing appearing in Italian should thus not possess argument and event structure, nor should the affix and base appear as productively combinable, distinct morphemes.

Here, it will be shown that contrary to this expectation, numerous occurrences with argument and event structure may be found in a large Italian web corpus for a sample of 81 English nominalizations ending in the suffix -ing that have been borrowed into (certain registers) of Italian. Section 2 summarizes previous research relating to morphological borrowing, with a particular focus on two different views of morphological borrowing developed by Seifart (2015) and Gardani (2021). Some background information on the morphosyntax of the English suffix -ing and on Italian action nominalizations will be introduced, and finally, I present a synopsis of existing work on the borrowing of English forms ending in -ing into Romance languages, including Italian. Section 3 describes how the corpus data were collected and processed. Section 4 gives a brief overview of the types of borrowings contained in the sample under analysis. Sections 5 and 6 next address the questions of whether the deverbal nominalizations ending in -ing in the sample present occurrences with argument structure and with complex event-structure readings. A brief conclusion explores some implications for our understanding of morphological borrowing and of the borrowing of verbal semantics more generally.

2 Previous research

3 Methods

4 Types of borrowings in the sample

Most nouns ending in -ing in the sample under study are pure loan words (in the sense of Winford 2005; Haugen 1950), i.e., words that imitate both form and meaning of the corresponding English lexical item. To be sure, many of the forms are polysemous and are used with a variety of readings in English as well as in Italian, such as shifting or booking. Given this high degree of polysemy, a close comparison of readings which are found in the SL and in the RL is beyond the scope of the present study; undoubtedly the meaning of a given noun in -ing is often not exactly the same in the SL and the RL. On the other hand, some of the words in the sample have a highly specialized semantics, such as kriging ‘method of interpolation in statistics’, which is imported quite exactly.

In collecting the data for this study, it became clear that the stems^[2] and the roots^[3] on which the forms in -ing are built occur in a variety of loan blends (in the sense of Haugen 1950; Winford 2005) in the corpus, i.e., in words that imitate only partially the form and meaning of elements from the SL, combining an element from the SL with an element from the RL. Since this study is concerned with the borrowing of deverbal nominalizations, it was decided to concentrate on loan blends which are verbs. With the aim of getting a more precise idea of whether for the 81 nouns ending in -ing studied here, the itTenTen16 corpus contains morphologically related verbs, I generated, for all 81 nouns in the list, firstly, hypothetical verb forms in -are (which is the unmarked inflectional class of Italian) that begin with the stem of the nouns ending in -ing (i.e., consisting of the form without the suffix -ing, e.g., prefetch- followed by -are – prefetchare – for prefetching) and, secondly, hypothetical verb forms that begin with the root of the nouns ending in -ing (i.e., consisting of the base without any affixes, e.g., fetch- followed by -are – fetchare – for prefetching) and then checked whether these hypothetical forms are attested in the corpus. Thirdly, I checked unsystematically for potential orthographical variants that I thought might exist, such as fecciare as a variant to fetchare, where the Italian grapheme <cci> is homophonous to the English grapheme <tch>. Finally, I also noted variants of morphologically related verbs that I had not systematically searched for, but which I came across while working with the corpus, such as networkizzare as a variant for networkare, hackerare as a variant for hackare or twitteare as a variant for twittare. This resulted in a list of 66 Italian verbs that are morphologically related to one of the 81 nouns in the list; see Appendix, Table 1. According to Table 1, there are only 26 nouns ending in -ing for which no morphologically related verb form, neither built on the stem nor on the root of the noun, was found in the corpus. For more than two thirds of the nouns in the sample, in contrast, a morphologically related verb, sharing at least some aspect of meaning with the noun, is attested.

Among the 81 nouns ending in -ing, one form may be classified as a loan blend, i.e., the neologism venertuning, which has a rather specialized meaning, referring to an event where persons show their tuned cars in public on a Friday evening. To date, this word continues to have a very specialized usage and it appears to be far from familiar to the common speaker of Italian.

That some Italian anglicisms ending in -ing are so-called false anglicisms has long been noted in the research literature (Gottlieb and Furiassi 2015; McLaughlin 2018; Renner and Fernández-Domínguez 2015). The Italian noun lowering illustrates this class. While the English noun lowering is a fully transparent nominalization of the causative verb lower, the Italian noun lowering is found in the corpus only as a piece of specialized terminology, pertaining to meteorology.

A surprising result of the present study is that we find not only pure loan words, but also a number of loan blends, and in particular verbs that share a stem or root with a noun ending in -ing; see Appendix, Table 1 for a list. These may be taken as evidence that speakers of those registers of Italian in which these forms are used have acquired some sort of representation of the morphological structure of these forms. To give an example, for speakers of Italian who use the noun prefetching as well as the verbs prefetchare and fetchare, it appears that their Italian mental lexicon contains a root fetch and a stem prefetch that have been added to the first conjugation class and are related by form and by meaning to a borrowed deverbal noun prefetching. A derivational suffix -ing (which, however, does not attach to native Italian bases) somehow connects the base verbs fetchare and prefetchare with the borrowed deverbal noun prefetching. Is this then a genuine case of borrowing on the level of constituent morphemes? Even though -ing, as suggested above, does not attach to native Italian bases and is thus not part of the native Italian noun morphology, a special subset of corpus examples indicate that, at least in this special pattern, some sort of morphological segmentation can be attested.

5 Do borrowed nouns ending in -ing have argument structure?

A dimension of semantics that is often considered as characteristic for verbs and that is much less relevant for morphologically underived nouns, is argument structure, (or valency in a different terminology), i.e., the level of semantico-syntactic representation on which information about the semantic roles that are assigned by a predicate is stored (Williams 1981). Verbs typically have at least one argument (e.g., laugh, which only assigns an agent role) but often more (e.g., give, which assigns three roles: agent, theme and goal or beneficiary), while prototypical nouns often do not have arguments to which they assign a semantic role (but in that case they do have a referential argument Baker [2003: 93]). Deverbal nominalizations, in contrast, have often been shown to inherit the argument structure of their base verbs (cf. Booij [1992], Williams [1987], Zwanenburg [1990] and much subsequent research). In previous studies on event structure and argument structure of deverbal nominalizations, it has been claimed that argument structure in nominalizations, and also complex event structure (see the discussion in Section 6) needs to be licensed by overt affixes (an assumption going back to Grimshaw [1990], Smith [1972] and which is taken up by, e.g., Borer [2014]). In fact, in studies on language contact, a similar restriction has been proposed: As early as 1975, Moravcsik (1975: 4) claimed,

that borrowed verbs, by internal syntactic composition, are (at least) bimorphemic and that they are bilingual, consisting of a generic verb constituent whose form is indigenous, and of a more specific nominal constituent whose phonetic form corresponds, by identity or similarity, to the phonetic form of the source verb.

In other words, verbs which are borrowed from a SL are integrated into the RL by adding some native functional element of the RL (which in Moravcsik’s view may be phonetically filled or unfilled, i.e., empty) that licenses what Moravcsik (1978: 111) terms verbal semantics, of which event structure and argument structure is one aspect (for a more fine-grained case study on argument structure in the borrowing of verbs, see Schirakowski, this issue). If Moravcsik’s claim is correct, then borrowed nouns in general are not expected to have argument structure (and complex event structure) – unless morphological borrowing in Gardani’s (2021) sense has occurred. Now, there is currently no evidence for morphological borrowing of the suffix -ing into Italian in Gardani’s sense since it does not attach to native bases. In this section, I will therefore address the question of whether the borrowed nominalizations in the sample occur in argument-realizing structures in the corpus.

Generally speaking, Italian nominalizations can realize their arguments in prepositional phrases headed by different prepositions (10a–b), as well as in a pronominal possessive determiner (10c), cf. Gaeta (2015) for an overview on Italian. In this respect, the Romance languages are quite similar (cf. Meinschaefer [2003] for French, Meinschaefer [2005a] for Spanish, Meinschaefer [2016] for a comparative Romance view), and they do not differ much from English (Rappaport 1983). To avoid manual inspection of thousands of occurrences from the corpus, I chose to focus on syntactic constructions in which the realization of the arguments of a deverbal noun is most likely; examples are given in (10).

The more than 50,000 occurrences of the 81 Italian -ing-nouns that had been extracted from the itTenTen16 corpus were thus searched with the help of the AntConc concordancer (Anthony 2022) for the sequences of word forms given in (11), which are the most common structures in which Italian deverbal nominalizations realize their arguments.

The searches resulted in 1,021 occurrences of Italian -ing-nouns with the relevant structures, which were exported from the corpus in spreadsheet format and then manually inspected and annotated for whether the respective prepositional phrases or possessive determiners in fact realize an argument of the event denoted by the Italian -ing-noun. After inspection, 163 occurrences were excluded because no arguments are realized; see (12) for examples. In Italian, not only the possessive determiner but also a prepositional phrase headed by di can have a possessive interpretation, as in (12a). Possessive interpretations are very rare among the occurrences under analysis, in which the noun in -ing refers to an event, but they are expected with nouns that have a concrete object reading, such as casing in (12a). Since most of the nouns ending in -ing refer to events in a broad sense (although they are not necessarily complex event nouns, see Section 6), temporal interpretations of the PP with di are far more frequent, as in (12c).

All in all, the concordances of the 81 nouns ending in -ing under study thus contain (at least) 858 contexts in which the noun has argument structure; see (13) for examples.

It is also evident, however, that only some of the Italian -ing nouns show argument structure in the corpus. For more than half of the Italian -ing nouns in the sample, none or only very few tokens appear with arguments; see Appendix, Table 1, for details. It remains to be studied which factors determine whether a noun ending in -ing can have argument structure or not. For our purposes here, the important point is that hearers and readers of these nouns are often being confronted with syntactic structures that can only be interpreted by attributing argument structure to the borrowed nouns. They thus acquire lexical items that are not simple pairings of sign and meaning, but have argument structure as well.

6 Do borrowed nouns ending in -ing have event structure?

A second semantic dimension which is typical for verbs is event structure. Verbs can refer to different kinds of events, among which we find processes that unfold in time but do not have a culminating point (e.g., boil), but also accomplishments, which likewise unfold in time until they reach a point of culmination (e.g. melt). Since previous studies have often claimed that complex event structure in deverbal nouns needs to be licensed by an overt affix (e.g., Borer 2014; Grimshaw 1990; Smith 1972) – which lacks in the -ing forms under study as long as no morphological borrowing has occurred – we will have a closer look at the eventive properties of the loan nominalizations in the sample.

The distinction between complex event nouns (CEN), simple event nouns (SEN), and result nouns (RN) was introduced by Grimshaw (1990: 45). To which of the three classes a given event noun pertains can be tested by means of various distributional diagnostics; see below. Only CEN share with their base verbs the property of taking arguments; (14a). CEN and SEN pattern together in that they refer to events, (14a–b), while RN refer to objects (14c). SEN and RN, on the other hand, pattern together in not having argument structure, (14b–c), in contrast to CEN, (14a).

The distinction between CEN and SEN has often been claimed to be associated with various distributional differences, such as the compatibility of the singular noun with certain modifiers like frequent or gradual, which modify the event structure, (15a–b) (cf. Melloni [2012: 25–27; 95–102] and Russo and Caselli [2021] for discussion of relevant criteria and their application to Italian, and Meinschaefer [2005b, 2016] for Romance languages more generally).

In a previous study on French and German loan nominalizations ending in -ing, Heinold (2009) has dealt with the question of how to describe the exact nature of the event-structural properties of these forms. She considers loan nominalizations ending in -ing in French and German contrastively, comparing them with the native English forms as well as with the native French suffixes -age and -ment and the native German suffixes -ung and -en (i.e., the nominalized infinitive). From studies on French nouns ending in -ing, Heinold (2009: 77) infers that in French this suffix can attach to native bases, hence, morphological borrowing seems to have occurred (see Section 2.1 above). For German, in contrast, Heinold (2009: 80) claims that -ing cannot attach to native bases, but this claim does not appear to be based on systematic data collection or previously published research. Heinold’s main point is that in French, nouns ending in -ing can have both durative and terminative aspect and refer to result states and objects (similar to the native suffix -ing in English),^[4] while in German they only can have terminative aspect and refer to results. Although Heinold’s (2009) concepts of durative and terminative aspect are slightly different – but they may be more precise (see Heinold 2009: 71) than Grimshaw’s concepts of Complex Event Nouns and Simple Event Nouns, – it appears that nominalizations with durative aspect are CEN although Heinold (2009: 76) claims (without showing that this is indeed the case by applying the relevant diagnostics) that French deverbal nouns ending in -ing cannot be CEN. Her point thus seems to support the general view that event structure needs to be licensed by an overt affix (e.g., Borer 2014; Grimshaw 1990; Smith 1972), which is indeed present in French loan words ending in -ing – since in French morphological borrowing has occurred (and nouns ending in -ing can have a durative reading) – but absent in German, where -ing is attested only in English loan words (and can have only a terminative reading). It is, however, not entirely clear whether Heinold’s description, and thus her more general claim, is correct as it is not based on any systematic data collection, neither on corpus data nor on a controlled acceptability study. Her German data are based on her own introspection (Heinold 2009: 79), while the French data for nouns ending in -ing are taken from Farge (2004) and refer to only a single distributional criterion, i.e., countability of the noun ending in -ing when preceded by faire ‘do’ (as in faire du camping, in which camping is a mass noun and preceded by a partitive, which Heinold therefore classifies as durative, and faire un briefing, in which briefing is a count noun and preceded by an indefinite article in the singular, which is therefore classified as terminative); cf. Meinschaefer (2005b) for a discussion of the relation between the countability of deverbal nouns and their status as CEN.

While Heinold’s claim is interesting and is compatible with the general view that complex event structure is available only in the presence of an overt affix, I will in a first step only be concerned with the question of whether nouns ending in -ing can have complex event structure at all – despite the fact that Italian -ing does not attach to native bases and that no morphological borrowing seems to have occurred according to Gardani (2021). As diagnostics I use the compatibility of nouns ending in -ing with event-modifying and with agentive adjectives in the present corpus. To be sure, a next step would consist in a closer inspection of their event-structural features in contrast to those of Italian nouns in -aggio, -tura, -zione or -mente, but this goes beyond the scope of this article.

To find contexts in which Italian -ing-nouns present complex event structure readings, I extracted concordances of -ing-nouns that co-occur with the event-modifying adjectives given in (16a), with the agentive adjectives listed in (16b), and with the eventive verbs listed in (16c).

The corpus search resulted in 234 occurrences, which were exported in spreadsheet format from the corpus. All occurrences were annotated manually for whether the Italian -ing-noun in fact has an eventive reading. After deletion of those examples in which a diagnostic was present in the larger context, but where the noun did not have an eventive reading, we were left with 135 occurrences of nouns ending in -ing with event structure; see (17) for examples.

The examples thus present clear indications that at least some nouns ending in -ing not only refer to events but also have complex event structure in the sense of Grimshaw (1990).

7 Discussion and concluding remarks

This study has shown that English deverbal nouns ending in -ing borrowed into Italian, such as screening, can have argument structure and complex event readings. Following Gardani’s (2021) observation that in the RL Italian the segment -ing is found only on bases from the SL English, this may be taken as evidence that morphological borrowing has not yet occurred so that the forms in question do not present an overt affix that could license argument structure and event structure. Should we therefore give up the assumption that argument and event structure need to be licensed by an overt affix, as proposed by Smith (1972), by Grimshaw (1990) or Borer (2014)?

Recall that a similar proposal was made by Moravcsik (1975: 4) with respect to the borrowing of what she terms “verbal meaning” (Moravcsik 1978: 111), and of which I assume argument and event structure are two aspects. On Moravcsik’s view, verbal meaning needs to be licensed in borrowed verbs by some (overt or covert) native functional element. While Moravcsik (1975) does not make entirely clear which elements of the RL can serve this function, it seems plausible that in borrowed verbs, inflectional affixes of the RL suffice to license verbal “meaning”, or structure; hence, a borrowed verb like screen-are can have argument and event structure because of its RL inflectional affixes (as in screen+ano ‘3.Pl.Prs’ or screen+avano ‘3.Pl.Pret’, as attested in the itTenTen16). Thus, the finding that these borrowed verbs show argument and event structure presents no contradiction to Moravcsik’s hypothesis: argument and event structure in screenare are licensed by the Italian verbal suffixes.

For the borrowed English roots like screen, many of which combine with native Italian inflectional affixes, the RL thus shows a “higher degree of entrenchment of the once-foreign elements” (Gardani 2021: 136) than for the borrowed element -ing. The question arises: what does Gardani’s “higher degree of entrenchment” actually consist of when considered from the perspective of the speaker? We believe that in the mental lexicon of a speaker of (at least certain registers of) Italian, who uses loan words like screenare and screenaggio, such words are represented as morphologically complex: they are segmented into root and affix, just like native Italian verbs. In other words, both the root and the affix must be mentally represented as distinct morphemes. In this context, it is also worthwhile to mention the concept of accessibility of borrowings as discussed by Winter-Froemel (this issue).

Now, what about the status of -ing in borrowed deverbal nouns like screening, whose roots are recognized as morphemes by (some speakers) of Italian? Research on morphological priming, which is one window into the mental lexicon of the speaker, has shown that a morphological segment (i.e., a segment for which morphological priming occurs) is not necessarily a genuine morpheme, combining both form and meaning. Again and again, it has been shown that morphological priming occurs not only for morphemes as segments with form and meaning, as in walk and walk-er, but also for pseudo-affixes, as in corn and corner (Marslen-Wilson et al. 1994; Taft 1981; cf. Frost et al. 2008: 934 for a brief review). From such findings we can conclude that segmentation of the formative -ing has certainly occurred in the mental lexicon (i.e., in the I-language) of (some) speakers of Italian, in the sense that this affix is mentally represented as a formative that can be separated from its root in processing. This finding allows us to understand why deverbal nouns can be borrowed with what Moravcsik has termed ‘verbal meaning’ and even exhibit argument structure and complex event readings: In the mental lexicon of the speaker, then, there is a morphological segment that, so it appears, fulfills the relevant function. However, we have no evidence that -ing has acquired the status of a derivational morpheme in Italian – differently from French and Spanish – since it combines only with English roots, producing words that are also found in English.

The reason for this contrast may be found in the different structures of the three Romance languages. Thus, French has lost all nominal class and gender markers (Klausenburger 1992) and Spanish has generally preserved them, with the exception of word-final /e/ after coronal consonants (Harris 1991, 1992). It is exactly to this form class that loan words ending in -ing are assigned, as in mitin, Sg – mítines, Pl (Mott 2015: 178). Italian, in contrast, has preserved nominal inflection across the board (Dressler and Thornton 1996: 5). Morphological integration of foreign affixes with the native lexicon is thus a more complex matter in Italian, an inflecting language, than in French, which has lost all nominal inflection, and Spanish, which has only one class with simplified morphological exponence. This, therefore, may very well be the reason for the resistance of speakers of Italian to combine -ing with roots from the RL Italian.