Németh T., Enikő: Implicit Subject and Direct Object Arguments in Hungarian Language use: Grammar and Pragmatics Interacting

Searching for missing arguments and the principles governing their distribution and interpretation within the depths of morphosyntactic, semantic and pragmatic theory is a quest filled with theoretical intricacies and empirical nuances.^[1] While such arguments have been amply studied, they continue to present a humbling research challenge, with each answer triggering more questions, Hydra style. Highly welcome are thus contributions such as Enikő Németh T.’s book, whose 4 core chapters (in addition to the introduction and the conclusion) aim at investigating the way in which ‘various grammatical (including lexical-semantic and morphosyntactic), general pragmatic, and particular contextual factors’ (p. 27) interact to make possible the licensing and interpretation of implicit subjects and direct objects. Accordingly, throughout the book the discussion focuses on the interpretive properties of the chosen types of implicit arguments in Hungarian and the formal, morphological and syntactic aspects are referenced to the extent to which they are relevant to interpretation.

The introductory first chapter situates the research within the author’s approach to the relation between grammar and pragmatics (they are separate ‘components of the theory of language, and an intensive interaction is assumed between them’ (p. 13)), embedding the discussion in methodological and metatheoretical considerations. It also outlines the empirical basis of the research, which consists of utterances as the units of analysis (rather than isolated sentences), coming from a variety of sources (the author’s 310-min long interview-based spoken corpus, the Hungarian National Corpus, dictionaries and other linguistics sources, native-speaker intuitions, minimal pair and thought experiments).

Reviewing different types of approaches to the study of implicit arguments in chapter 2, that is the purely syntactic, pragmatic, and lexical-semantic approaches, as well as approaches combining these levels of analysis, Németh T. argues that the former are insufficient, and it is only the final, complex type of an approach which can explain the patterns of behaviour of verbs with implicit arguments.

By presenting the definitional considerations which enter into the research on implicit arguments and by outlining ‘the manners of occurrence of implicit arguments in Hungarian’ (p. 28 and thereafter), chapter 3 sets the background for the study of the lexical semantics of verbs with implicit arguments (chapter 4) and the role of contextual and encyclopedic information in their licensing and interpretation (chapter 5). Implicit arguments are defined as ‘arguments involved in the lexical-semantic representations of verbs, but which are lexically unrealized, and whose implicit presence in utterances is attested by lexical-semantic, grammatical (phonological, morphological, syntactic, and semantic), discourse, and/or pragmatic evidence’ (p. 67). Thus, in addition to phonemically null definite and indefinite subjects and objects, under this definition fall also, for example, the expressions of location with verbs such as rain (e.g., It is raining (here)) and instrument with verbs such as kill (Mary killed John (with a gun)), though the focus in the discussion of Hungarian is on subjects and objects, as signalized by the book’s title.

The core argumentation, presented in chapters 4 and 5, is structured around the division of implicit arguments into three ‘manners of occurrences’, where the distinguishing factor is the type of information making it possible for the meaning of the argument to be recovered: manner (A) the verb’s lexical semantics; manner (B) typical interpretation triggered by the immediate utterance context, including encyclopedic information and grammatical requirements; and manner (C) extended context. All of these operate according to Sperber and Wilson’s (1986) cognitive principle of relevance. All Hungarian verbs can be used with implicit arguments in manner (C), which seems to follow from the pro-drop nature of the language. Some verbs can do so in manner (B) in addition (e.g., ad ‘give’, as in Ági (pénzt) adott a koldusnak ‘Ági gave (money) to the beggar’ (p. 74)), and yet some others can follow all three patterns (e.g., eszik ‘eat’). These three manners guide the process of interpretation of sentences with implicit arguments in actual language use, as Németh T. posits that the hearer ‘attempts to form an adequate interpretation in a particular context starting from a lexical-semantic representation of the verb’ (p. 75) and with this failing, manners (B) and (C) are consulted consecutively. Though an interesting idea against the background of the implicational relation between the three manners (if a verb can be used in manner (A), it can be used in manner (B) and (C) and if it can be used in manner (B), it can also be used in manner (C)), one may wonder whether such a linear process of interpretation indeed can be supported by empirical or more detailed theoretical arguments favoring it over, say, a rapid integration of all relevant pieces of information as they become available in parsing. After all, as all verbs can appear with implicit arguments in manner (C) and only a small proportion of these is in addition consistent with manner (A), it could be expected that the majority of instances when the hearer needs to interpret an implicit argument in Hungarian requires appeal to context. Demanding that the hearer always consult the lexical semantics of the verb first anyway may thus not be the most economical solution, especially that it can work only for indefinite arguments, mostly in the object position.

The discussion of implicit arguments in the context of the lexical-semantic representation of verbs in chapter 4 begins with argumentation aimed at showing that verbs of natural phenomena (including, e.g., weather predicates), which can in some contexts be used with lexicalized subjects in Hungarian, encode subject arguments in their lexical-semantic representation, unlike what has been assumed in Hungarian grammatical tradition, which analyses them as subjectless. Németh T. draws an interesting parallel between implicit subjects of such verbs (i.e., e.g., hajnalodik ‘[for day to] break’, havazik ‘snow’) and implicit objects of English verbs of bodily emission and of contribution (e.g., vomit, donate), proposing that, just as has been suggested in some previous literature for the latter, also in the former case the relevant argument is present in the lexical-semantic representation of the verbs and it is identified based on their very specific selectional restrictions (p. 88–89).^[2] The same solution is suggested for verbs of work such as arat ‘harvest’ and takarít ‘clean’, in which case the implicit subject argument in sentences such as Poros az asztalom, tegnap se takatítottak rendesen ‘My desk is dusty, yesterday [the cleaners] did not clean [the office] well’ is interpretable owing to the selectional restrictions of the relevant verbs, which are taken to specify the agent argument (e.g., takarít ‘clean’: ‘action of x who is a cleaner causes y to become clean’ (p. 105)). However, in this case it is unclear why sentences in which the explicit subject is ostensively not a cleaner are completely unremarkable (e.g., A tanár takarítja az asztalát ‘The teacher is cleaning her/his desk’). A solution implied in the discussion is that context can override selectional restrictions (p. 110), but when one compares the ease in which this would have to happen in the cases at hand with what is observed in other instances of interpretation of clauses where on standard assumptions the overt argument indeed violates the verb’s selectional restrictions (e.g., the object in Anna is drinking some stones), the need to bend the imagination in the latter but not the former case seems problematic. What is more, at least impressionistically, on everyday basis verbs such as takarít ‘clean’ seem to be used much more frequently with agents who are not professional cleaners etc., which raises additional doubts, both with respect to the economy of linguistic representation and the acquisition of such entries by children.

The second part of chapter 4 tackles the issue of implicit direct objects, where the interesting discussion of the interaction of lexical-semantic and pragmatic information embeds the relevant familiar empirical observations about the interpretive properties of indefinite implicit objects of verbs such as eat and drink, and their Hungarian counterparts. One of the intriguing facts which can inspire further, perhaps comparative research is that in Hungarian not only the verbs eszik ‘eat’ and iszik ‘drink’ can be used with unspecified object drop, but so can the verbs of manner of ingestion, that is zabál ‘devour’ and vedel ‘swill’ (p. 134). This is of course different than what we find in English, where the latter are much more resistant to object drop and make it possible only under some specific contextual conditions.

In chapter 5 the discussion shifts towards the interpretation of implicit arguments based on immediate utterance context (manner (B)) and extended context (manner (C)) in Hungarian, a consistent pro-drop language, where not only subject pronouns, but in some cases also object pronouns can be dropped. The default interpretation of zero subject anaphors as co-referential with the subject/topic of the preceding clause is shown to be sometimes overridden by pragmatic considerations. As far as objects are concerned, an interesting feature of the Hungarian system is the distinction between the definite and indefinite conjugation, where the different markings on the verb can support the identification of the referent as the first or second person, as opposed to the third person, in that the former features on the object (be it null or overt) require the use of the indefinite conjugation. In contrast to some previous claims in the literature, Németh T. shows that, provided the right type of context, not only singular, but also plural first and second person objects can be dropped, a finding also reported in Keresztes (2013). Furthermore, she provides a number of examples with third person plural null objects, both in anaphoric and exophoric contexts. That third person plural object drop is available in Hungarian at least in some environments is a particularly exciting empirical observation, as it contradicts all previous literature on the topic and so can lead to further research aimed at elucidating the exact nature of the contrasts observed in this domain.

Another context discussed are cases where the implicit object argument is interpreted with reference to encyclopedic knowledge typically associated with different elements of the utterance (e.g., in a sentence containing an implicit object of the verb ad ‘give’, the agent, and a dative phrase referring to a beggar, the object is interpreted as money in this way (p. 176)). The final manner of implicit argument interpretation (manner (C)), which relies on the extension of the immediate utterance context with information from the discourse, the physical environment, and further encyclopedic knowledge, is the only one which is consistent with the verb inflected in accordance with the definite paradigm.

Generally speaking, the discussion in the book is oriented towards showcasing the different routes through which implicit arguments are licensed and interpreted in Hungarian. This aim is achieved, but as my own research focuses mostly on syntax and its interfaces, I feel compelled to note here that from a syntactician’s perspective, there are a couple of problematic issues which arise in the course of the discussion. Firstly, while it is clear that explaining all aspects of implicit arguments in linguistic competence and in language use requires reference to lexical, grammatical, and contextual factors, the claim that ‘the encyclopedic information as well as particular contextual factors can override the use or interpretation predicted by the grammar’ (p. 66 and thereafter) is rather controversial. After all, in the practice of constructing theoretical models, when one encounters data falsifying an aspect of the postulated system, the model is revised rather than maintained and patched with stipulations at a different level of representation. Simply put, if a reading inconsistent with the postulated grammar is available, the grammar needs to be revised. This is at least how the history of research within the generative theory has proceeded. Other approaches are certainly possible, but since this issue relates to some fundamental properties of the architecture of the human language, it requires explicit discussion, including a hypothesis about what is special about the contexts where overriding grammatical rules is possible, as opposed to the structures where it is not.

Secondly, the syntactic literature reviewed is not very comprehensive or up to date. This is certainly understandable given the focus of the work and the vastness of the available syntactic sources, but it is somewhat problematic for the critical discussion of ‘purely syntactic explanations’ in Section 2.1, where it is unclear whose exact approaches are criticised. For example, it is stated that ‘purely syntactic approaches are not sensitive to contextual analyses and they cannot take into consideration implicit arguments in utterances’ (p. 30), but there are no particular analyses mentioned to which this point would apply, other than a reference to Hungarian grammatical traditions (represented by Komlósy 1992, 1994), where the main point is that sentences such as Áron tologat ø ‘Áron is pushing ø back and forth’ (p. 31) are actually acceptable when embedded in appropriate context. However, this is not very surprising and it is hard for me to think of a theoretical syntactic analysis of any type of elliptical structures which would not take linguistic and extra-linguistic context into account when presenting acceptability judgments, which are always produced for a given string under a relevant interpretation.

Syntacticians may also find surprising the suggestion that the existence of so-called discourse pro-drop languages such as Mandarin, where the licensing and identification of null subjects is achieved with reference to discourse entities, is incompatible with generative grammatical approaches (p. 32–33). After all, this empirical observation entered the theoretical debate through Huang’s certainly generative 1982 dissertation and a number of generative studies on null arguments make direct reference to notions such as topic and the speaker and hearer features (see, a.o., Frascarelli 2007; Huang 1984; Raposo 1986; Sigurðsson 2011).

Despite the abovementioned issues, the book is a splendid resource for anyone interested in implicit arguments in Hungarian and beyond. By providing rich empirical material exemplifying a variety of implicit arguments rather than focusing narrowly on a chosen type, it offers a refreshing bird’s eye view on the mechanisms governing their interpretation through considering the interaction of different components of linguistic knowledge as applied in actual language use. As the discussion is structured around the very clear division into three manners of implicit argument interpretation, it will be useful not only to researchers, but also to advanced students just setting out on the quest in search of missing arguments.