Manner expressions in Finnish and Estonian: their use in quotative constructions and beyond

1 Introduction

Recent crosslinguistic studies show that manner expressions (MEs) like comparative/similative markers with the meaning ‘like’ and demonstrative deictics ‘so’ can conventionalize as quotative markers (see Buchstaller and van Alphen 2012: xii–xiv; Güldemann 2008: Section 5.1.3–5.1.4). A similar development is observed in the two Uralic languages from the Finnic branch Finnish and Estonian (see Teptiuk 2019a, 2019b, 2020). Similative adpositions (Fin. niin kuin (niinku), tyyliin, Est. nagu, a la, all ‘like’) and manner demonstratives (Fin. sillee(n), Est. nii ‘so’) become conventionalized as quote-introducers. It means that they can form a quotative index (QI), i.e., a segmentally discrete linguistic expression indicating the presence of reported discourse (Güldemann 2008: 11), either co-occurring with reportative verbs like ‘say’ and ‘think’, or with the non-reportative equational verb ‘be’, similarly to the Eng. QIs be like/all. The conventionalization of similative prepositions goes even further, and they are sometimes used as single quote-introducers (Teptiuk 2019a: 216, 223). Consider (1) illustrating the use of the similative preposition niin kuin (niinku) ’like; as if’ in QIs in colloquial Finnish. Reported discourse in examples is enclosed in double vertical bars.

Similative prepositions in QIs contribute to the presentation of reported discourse. They present reported utterances and thoughts as approximately reproduced (1a)–(1b) or fictitious (1c) (Teptiuk 2019b: 291–293). As for manner demonstratives, their use is primarily associated with cataphoric reference, i.e., pointing at the following stretches of reported discourse (Teptiuk 2020: 284). Besides reported utterances and thoughts, some markers like the Finnish distal manner demonstrative silleen ‘thus; that way’ can introduce non-verbal demonstrations, such as an emoticon in (2).

Although it is hypothesized that cognitive verbs behave similarly to reportative verbs in complex sentences (see e.g., Güldemann 2008: 4; Palmer 1986: 135), the use of MEs with epistemic (‘know’, ‘guess’) or perceptive (‘see’, ‘hear’) verbs has not been investigated so far in the two Finnic languages. In this study, we expand previous accounts and investigate the use of MEs previously observed in quotative constructions for their combination with epistemic and perceptive verbs and check if they show any restriction here. In addition, we check for functions expressed by MEs in combination with these types of verbs, introducing propositional or illocutionary content (see Boye 2012: Section 4.2.1; Davidse and Vandelanotte 2011: 241; Holvoet 2018).

We target three types of perceptive verbs: (i) visual with the meanings ‘see’, ‘be seen’/‘seem’; (ii) auditory: ‘hear’, ‘be heard’/‘sound’, and (iii) sensory: ‘feel’, ‘be felt’, and three epistemic verbs ‘know’, ‘understand’, and ‘guess’. Our choice of verbs is not accidental. Reports of perception are considered “inherently more similative than reports of speech” (Meyerhoff 2002: 352). Hence, one would expect to find the combination of perceptive verbs with MEs quite naturally in contexts where speakers evoke additional connotations to what they report to hear or see. In the case of epistemic verbs, we hypothesize that the combination with MEs contributes to the expression of propositions similarly to MEs in quotative constructions. The co-occurrence of manner demonstratives with the epistemic verb ‘know’ should mostly reflect the endophoric use of demonstratives. With semantically more subjective verbs like ‘guess’ and ‘understand’, such a combination might highlight the subjective interpretation of events. As for similative markers, instead of asserting the proposition denoting epistemic processes, the speaker might present them as approximate, non- or counterfactual (see Section 2). With the verb ‘know’, this combination may lower the speaker’s epistemic support and express distancing in contrast to the presupposition of the proposition to be true, found in cases without MEs.

The article is organized as follows. Before turning to the presentation of methodology and data in Section 3, we describe perceptive and epistemic verbs together with MEs in Finnish and Estonian in Section 2. The results are discussed in Sections 4 and 5. The discussion is organized according to the type of ME. First, we look into the combination of the verbs with manner demonstratives (Section 4) and then with similative prepositions (Section 5). Finally, Section 6 summarizes the results and discusses functional similarities in Finno-Ugric languages spoken in Russia. Similarities outside the Finnic branch are used to generate further hypotheses about the crosslinguistic use of MEs in complex sentences.

2 Epistemic and perceptive verbs and manner expressions in Finnish and Estonian

Two types of verbs are investigated in this study: epistemic and perceptive. These verbs are presented in Table 1. This is not an exhaustive list and can be supplemented by verbs that derive the targeted meanings in particular contexts as semantic extensions of their primary meanings. Such verbs are also present to a certain extent in Table 1. For instance, the Est. verb arvama has the primary meaning ‘think’. The meaning ‘guess’ appears in contexts where the process of thinking is directed at inferring other people’s thoughts (e.g., ‘I tried to guess [i.e., think] what he meant/thought about’). In addition, it may signal the speaker’s uncertainty (‘I don’t know, but I guess so’) or mistaken belief (e.g., ‘I guessed so, but it turned out differently’) (cf. Õim 2007).^[2]

Another such verb is Finnish tuntea with the primary meaning ‘be acquainted’. The meaning of sensory perception ‘feel’ is only secondary for this verb. We decided to investigate tuntea mainly due to its etymological relatedness to the verb tuntua ‘feel’ used more conventionally in the expression of feelings. By investigating this pair of cognates, we could check if there is any visible difference in using these verbs with MEs in complex sentences with reports of perception or epistemic processes.

Perceptive verbs in Table 1 are represented by pairs of verbs: transitive and intransitive. In all cases, except for Est. kõlama and paistma, the verbs forming a pair are obvious lexical cognates.^[3] Transitive verbs precede intransitive in the table. The experiencer is the subject of transitive verbs but not of intransitive verbs where it is marked with the allative case (highlighted), e.g., Fin. minu- lle tuntuu, Est. minu- le tundub, lit. ‘(it) feels to me’.^[4]

The translations of perceptive verbs in Table 1 show that some have more complex meanings. For instance, visual verbs can fall on the borderline between visual and inferential verbs; sensory verbs– on the borderline between sensory and inferential (in both cases, reflected in the translation ‘seem, appear’). In this study, we take an inclusive approach and investigate the different meanings of these verbs. However, it is worth specifying that we are reserved in our expectations finding all meanings equally represented. Furthermore, it might not always be possible to distinguish between the meanings. Therefore, we primarily rely on the content that these verbs introduce and the context in which they appear to pinpoint the most appropriate interpretation of the verb in the complex sentence.

The system of MEs is no less complex. Among the category of demonstratives, Finnish distinguishes between speaker-proximal, hearer-proximal and distal manner demonstratives, marked by glosses next to translation in Table 2. The same distinction is missing to a large extent in Estonian.^[5] The distal counterpart naa is used only in contrast to proximal nii that is otherwise referentially neutral. Due to its restricted use, naa is not included in our investigation. The Estonian manner demonstratives in Table 2 are synonymous; nõnda is mainly used in colloquial speech. The same exclusively colloquial use is mentioned for the Fin. demonstratives tälleen, tuolleen and silleen (cf. Hakulinen et al. 2004: Section 721).

The similative prepositions in Table 2 can be roughly translated as ‘like, as if’.^[6] A close correspondence can be pointed out on the one hand between Fin. niin kuin and Est. nagu, justkui, and on the other, between Fin. tyyliin and Est. a la. The first group can be used in similative/comparative contexts (‘like’), as well as marking non- or counterfactuality (‘as if’). Tyyliin and a la, in turn, are used in similative/comparative contexts only.^[7] Thus, one of our tasks would be to determine if there are any differences between the use of these similative markers in complex sentences with perceptive and epistemic verbs.

3 Methodology and data

To check our hypotheses, we conducted a pilot corpus-based study. We used data from social network sites (SNS) as a database. The following principles governed the choice of material. First, SNSs can be conveniently used as a database of non-standard written communications, available in free access online. The significant advantage of such a database is a constantly growing corpus of present-day language use, which enables us to grasp contemporary processes in the language due to (i) the general lack of puristic tendencies among online communicants (see Pischlöger 2014, 2016) and (ii) dominance of spoken language in written form.^[8] Second, our preference for substandard written communications is also motivated by the lack of any limitations to specific sociological characteristics of speakers (e.g., age, social status, gender, dialectal background) and genre- or register-related restrictions that could lower the possibility of co-occurrences investigated here. The language on SNS per se can be considered a written approximation of spoken language, combining the features of colloquial speech and standard writing in one text (see below).

For Finnish, we used the database of internet communications (Fin. Internet-keskusteluaineistoja, henceforth IKA) accessed via the platform Korp.^[9] The corpus contains ca. 621 million sentences with ca. 7 billion words. For Estonian, the corpora etTenTen 2013, 2017 and 2019 were used via Sketchengine.^[10] The size of the corpora is ca. 116.5 million sentences and 1.5 billion words. The corpora include texts from various websites, but only SNS texts (forums, blogs, commentary sections) were used for this study, leaving out e.g., newspaper articles and stenogram texts from parliament meetings. Since some queries did not provide sufficient data, Google Search (GS) was additionally used.

This study is a part of a more extensive investigation, including three Finno-Ugric languages spoken in Russia: Komi (Permic branch), Udmurt (Permic branch), and Erzya (Mordvinic branch). Since data from these languages are also discussed in Section 6, a couple of words are in order here about the corpora used for Komi, Udmurt, and Erzya. Our material for these languages derives from the social media corpora of the Corpora of Uralic Volga-Kama languages.^[11] We exclude these languages from the main body of this article because of the unsystematic representation of co-occurrences and small amount of examples in the corpora. To avoid making hasty generalizations about the use of MEs in complex sentences with perceptive and epistemic verbs in these languages, we decided to continue our investigation of these languages outside this study using a different data source for them in the future. However, some similarities in the use of MEs are noted in Section 6 to generate hypotheses for further studies on this topic among Finno-Ugric languages and beyond.

The corpora and GS were used with the same principle, keeping in mind the differences between these two databases. The corpora allow a more convenient manipulation with the query. Therefore, the query was designed as follows: lemma of the verb + ME, e.g., nähdä ‘see:inf’ + näin ‘so.prox.sp’. If the first one hundred results did not offer sufficient or relevant results, the complementizer (Fin. että, Est. et) was added to the query. Such manipulation would allow limiting the number of examples to those containing clausal units following the complementizer and would likely involve examples relevant for this study.

For the Estonian material, the control mechanism was used when the corpora did not provide sufficient data and contradicted Eda-Riin Tuuling’s native-speaker intuition. According to her native-speaker intuition and depending on the verb, additional elements were added to the etTenTen queries, e.g., näen seda asja nii ‘I see this (thing) so’. If such a method did not improve the search results, we additionally used GS. GS does not allow the use of lemmas. Therefore, the query was constructed using the limited number of forms: prs.1sg, pst.1sg, cond.(3sg),^[12] cond.1sg, inf, pp.

The number of results produced by a GS could be too overwhelming; therefore, we limited the results the following way. First, we put the query in quotation marks to search for the whole phrase. Second, we added the noun foorum ‘forum’ to limit the sources, e.g., “näen nagu et”^[13] foorum. We looked at the first 20 search results, although the number was bigger for only a few queries. The authors acknowledge that these search tactics might leave out some potential results, e.g., sentences without the complementizer. However, the implemented queries should suffice for a pilot corpus-based study and some initial generalizations.

In this study, we took a semasiological approach, i.e., moving from form to function. We largely considered additional contextual and grammatical cues while formulating our interpretations and classifying the examples according to the types of meanings that the whole construction with ME denotes. All examples are provided here with translation and glosses in their original form, i.e., excluding corrections. Examples from Finno-Ugric languages spoken in Russia are presented in transcription without the original text in Cyrillic.

We supplement our observations with primary quantitative data listing the number of examples for each strategy in our material. Longer variants of manner demonstratives in Estonian nii/viisi, -moodi, -pidi, etc. (see Table 2) are not separately mentioned and are counted as a variant of the shorter form of the same manner demonstrative. We refer to these forms in the text only where the longer forms were more prominent than their shorter counterparts. The numbers provided here only suggest possible associations with and between the different strategies discussed here. A more fine-grained corpus study will be necessary in the future to provide robust quantitative outcomes.

4 Manner demonstratives in complex sentences with perceptive and epistemic verbs

In this section, combinations of manner demonstratives with perceptive and epistemic verbs are discussed. The possilbe combinations are summarized in Table 3. In this table, we refer to the meanings expressed by the verbs rather than naming them (for more details, see Table 1). Only the manner demonstratives observed in such combinations are mentioned.

Finnish hearer-proximal demonstratives noin and tuolleen are only marginally used with epistemic and perceptive verbs. There is no evidence of adjacent combinations with noin or tuolleen for the verbs ‘know’, ‘guess’, ‘be seen’, ‘hear’, and ‘be heard’, where this combination would introduce clausal units. For tuolleen, there is no evidence of co-occurrence with the verbs ‘understand’ and ‘see’. However, it was observed in combination with the verb tuntea ‘be acquainted; feel’ introducing the description of sensory perception. As for other verbs, hearer-proximal demonstratives are used in similar contexts and carry out the same functions as the rest of the demonstratives in combinations with particular verbs.

In Estonian, the basic manner demonstrative nii is mainly prevalent in our data and co-occurs with all verbs. Most verbs also appear with other manner demonstratives investigated here and do not seem to reflect any restrictions. Although some co-occurrences were not found in the data, according to Eda-Riin Tuuling’s intuition, they would still be possible and require further investigations using different data sources.

The following four functions were observed in the use of manner demonstratives: (i) mere endophoric, (ii) subjective interpretation of the event, (iii) reportative, and (iv) wishful thinking. In the remainder of this section, we will look at these functions in separate subsections.

5 Similative prepositions in complex sentences with perceptive and epistemic verbs

This section deals with similative prepositions used with perceptive and epistemic verbs. Table 8 is analogous to Table 3, listing the meanings of the verbs and similatives used with them in our data.

Not all similatives are used equally with perceptive and epistemic verbs. As hinted in Section 2, there is a division of labor between similatives. Finnish tyyliin and Est. a la, both ‘like’, mainly introduce reports, regardless of the verb they co-occur with. In contrast, Fin. niin kuin and Est. justkui, both ‘like, as if’, predominantly express epistemic overtones or mark propositions as counterfactual. Est. nagu does not show preference to a specific function and can introduce report, as well as fulfill other functions in complex sentences.

The following four types of functions were observed during our investigation: (i) reportative use, (ii) epistemic evaluation, (iii) subjective interpretation of the event, and (iv) marking proposition as counterfactual. The remainder of this section is dedicated to these four functions discussed in separate subsections.

6 Discussion and outlook

In this section, we will summarize the functions observed in the use of MEs in Finnish and Estonian and present them parallel to our results from other Finno-Ugric languages from our initial survey on five languages (see Section 3).

The first function observed for manner demonstratives is endophoric where the manner demonstrative simply points to the following presentation of the proposition in the complex sentence. Such a function is present in combination with the visual perceptive verbs ‘see(m)’ and can also be observed among Erzya (Mordvinic), Komi and Udmurt (both: Permic). Similarly to the two Finnic languages, the manner demonstrative points to the proposition without contributing to its expression. For instance, the Komi speaker reports how she saw the word in the dictionary in (27).

Another function observed among manner demonstratives is the subjective interpretation of the event. The combination with manner demonstratives resulting in such a funciton is quite harmonic with inherently subjective verbs like ‘understand’ and ‘guess’. As for the combinations with the verb ‘know’, it forces non-factive interpretation of the verb and results in non-factive construal describing the way of knowing specific to the reported speaker. In Estonian, the manner demonstratives combined with ‘know’ may also serve as a softening politeness strategy lowering the speaker’s commitment to the proposition’s truth value in self-reports of knowledge. In addition, visual and auditory perceptive verbs may acquire the inferential meaning and indicate that the proposition results from the speaker’s interpretation of some situation, lacking assertiveness. Among Finno-Ugric languages spoken in Russia, the same function could be observed in combination with visual ‘see(m)’ and epistemic ‘understand’. Consider (28) from Udmurt where the speaker presents their stance.^[17]

The reportative function is naturally observed in the combination of manner demonstratives with auditory perceptive verbs ‘hear’ and ‘be heard’. The same holds for Finno-Ugric languages spoken in Russia, as (29) from Erzya illustrates. Note that in this example, the speaker hedges from the report of rumors. Thus, in some cases, the manner demonstrative can also add to the epistemic evaluation of the report. However, the epistemic meaning expressed by manner demonstratives is context dependent and was not observed in each case where the manner demonstrative introduced reports.

The only function we do not observe among Finno-Ugric languages spoken in Russia, but which we have traced in Finnish and Estonian, is wishful thinking. This function is observed in constructions with visual perceptive verbs in the conditional mood. Due to the lack of sufficient examples in our corpus of Finno-Ugric languages spoken in Russia, it is too early to exclude a possibility of a similar use of visual perceptive verbs and further investigation is needed.

Now let us turn to the similative markers. The first function discussed for similatives is reportative. The conventionalized quotatives tyyliin in Finnish and a la in Estonian do not necessarily co-occur with auditory perceptive verbs, as is the case for manner demonstratives (see above) and the similatives niin kuin in Finnish and nagu in Estonian. The similatives tyyliin and a la usually introduce habitual quotations, typical for a concrete speaker, group of people, or a situation; niin kuin and nagu mark reports as approximately reproduced. The situation is partially similar in Udmurt and Erzya; our Komi data do not reflect the reportative use of similatives. In Udmurt, we observe the use of the Russian. similative tipa with auditory perceptive verbs, where this combination introduces approximately reproduced quotations (30a). In Erzya, the similatives can mark reports as counterfactual (30b) or doubtful on behalf of the reporter (30c).

In (30b), the reporter demonstrates how the reported addressees wrongly perceived Erzya paro ‘nice’ for Russian paru ‘more steam’. Hence, the similative ťeke ‘as if’ highlights the counterfactuality of this quotation. In (30c), the reporter expresses their doubt concerning the reported information that an artifact discussed might be of the Erzya origin, marked with the similative buto (<Ru. budto)^[18] ‘as if’.

The epistemic evaluation is also observed among similatives when combined with the epistemic verb ‘know’. Besides Finnish and Estonian, such use can be observed in Udmurt; other languages hint at the non-harmonic combination of ‘know’ with similatives. Note that this combination is not entirely harmonic in our Finnish data, where such cases are also scarce, unless they mark habitual quotations, as in (19). Example (31) demonstrates such use in Udmurt where the clause-final similative kaď highlights the speaker’s uncertainty towards their knowledge.

In addition to epistemic evaluation, similatives may indicate the subjectivity of the speaker’s perception. In Finnish, the subjective interpretation of the event is observed with the sensory perceptive verbs that acquire the inferential reading. In Estonian, such a meaning can be observed in combinations of similatives with all verbs investigated here. In Udmurt, we observe the combination of clause-final kaď ‘like’ with epistemic ‘understand’ and visual ‘see(m)’ (32). In another Permic language Komi, only the combination of the similative byťťökö ‘as if’ (<Ru. budto by) with visual kažitćyny ‘seem’ was observed.

The final function observed among similatives is counterfactual marking of propositions. In Finnish and Estonian, we observed such cases with visual and sensory perceptive verbs. Often such a function can be observed among exaggerated descriptions of visions and inferences. The counterfactual reading can be contextually specified by explicitly pointing to the scenario opposite that described in the proposition introduced by the verb and the similative. In Finno-Ugric languages spoken in Russia, we mainly observe such cases among visual and sensory perceptive verbs; however, (30b) illustrates such function in the quotative uses of similative ťeke in Erzya. Similarly to Finnish and Estonian, when the counterfactual proposition is introduced, the speaker highlights that they know it is not the case. Consider (33) from Udmurt where the speaker indicates with the Russian. similative budto that the feelings described are counterfactual, and the interlocutor discussed in the proposition is not from Kyrgyzstan.

As cases from two Finnic and three Finno-Ugric languages of Russia show, we can observe a similar behavior of MEs in complex sentences. Based on this pilot study, we can conclude that there is a functional correspondence in the behavior of MEs in reports of auditory, visual, and sensitive perception and the epistemic processes of knowing, understanding, and guessing across languages. Our classification of functions highlights four functions for manner demonstratives and four for similatives. However, it is still tentative and should be supplemented by investigations among different data sources.

As some examples show, the idiosyncratic behavior of MEs in complex sentences can also be expected in concrete languages. For instance, recall the combination of similatives with ‘know’. Such combinations are quite harmonic in Estonian, but Finnish and Udmurt data show only a few cases of such co-occurrence, while Komi and Erzya indicate a complete lack of examples in our data. Furthermore, some cases discussed here demonstrate overlaps between different meanings expressed by MEs.

For instance, subjective interpretation of the event and approximate presentation of quotations can be largely characterized by epistemic distancing. In some cases, manner demonstratives used endophorically result in a politeness strategy where the speaker lowers the commitment to the proposition’s truth value and indicates the contrast between their knowledge and other speakers’ reports. Such cases are usually distinct from epistemic evaluation, where the speaker explicitly highlights their uncertainty towards the proposition presented, although at times they appear very close. The epistemic evaluation can also be observed among other uses of MEs, e.g., in quotative constructions as (30c). Hence, some examples could equally fit into a couple of subsections discussing different functions here.

Furthermore, what also complicates the classification attempted here is the role of context in arriving at a concrete interpretation and polyfunctionality of some verbs. As some counterfactual cases show, their interpretation is based solely on the context contrasting two different events, without which it would be impossible to distinguish them from other cases. Thus, counterfactual reports, as in (30b), are likely to be interpreted as approximately reproduced without taking the context into account.

The polyfunctionality of some verbs allows them to take part in framing inferential processes, as in the case of sensory perceptive ‘feel’ in Finnish and Estonian. In turn, such uses with MEs often highlight the subjective interpretation of the event, resulting from inferences made by the speaker. Other verbs like visual ‘see’, marked with conditional and co-occurring with manner demonstratives, can express the speaker’s wishes. Thus, a large part of the meaning is already expressed by the verb itself and can be additionally supplemented by the pragmatics of situation described.

One way to improve such classification would be to look at such combinations on the level of construction and design the typology of functions starting from the verbs and regardless of the type of ME. Such a direction could also include other types of clause linkers, e.g., epistemically neutral complementizers, manner question words (‘how’), and manner nouns (‘way’), and compare their use with similatives and manner demonstratives, outlined here.

Although our classification could be improved, it provides first accounts on the use of MEs in complex sentences outside the quotative domain in two Finnic languages, which sets the foundation for the research on the same topic among other Finno-Ugric languages and beyond. This article also expands previous accounts on using these categories in quotative constructions and describes additional functions fulfilled by similatives and manner demonstratives in complex sentences.