Log in / Register
Home arrow Language & Literature arrow Theory and Data in Cognitive Linguistics

Comparison of the semantics of the Dative Subject Construction in Old Norse-Icelandic, Archaic/Classical Latin, Ancient Greek, Old Russian, and Old Lithuanian

Our database with dative-subject predicates has yielded 232 different sememes across Old Norse-Icelandic, Early and Classical Latin, Ancient Greek, Old Russian and Old Lithuanian, which we list in the Appendix together with one predicate from each of the languages where a verb with that meaning is attested. The term sememe refers to a unit of meaning, in our case the meaning of individual predicates. Several of the verbs are synonymous or near-synonymous, in which case we list only one predicate as an example of that sememe. This means that the Appendix does not list all the predicates in our database, but only one lexical instantiation of each sememe in the languages where it is attested.

Table 1 gives the type frequency of the dative-subject predicates found in each of the five Indo-European language branches. The predicates have been extracted through searches in dictionaries and reference grammars for each language, both electronically and manually. Some text-specific searches have also been made. The data from Latin were partially extracted from a computational lexicon containing morphological and syntactic information about arguments of Latin predicates occurring in two syntactically annotated corpora of Latin (see McGillivray 2010, 2014).

These predicates are well known from the philological and school grammar traditions as having an oblique argument which is not regarded as being the grammatical subject but the "logical" or the "psychological" subject. The argument structure and the meaning of the predicate together have been used to distinguish these predicates from other predicates where the oblique argument is syntactically an object. We are currently preparing an electronic version of our database (available at which includes the full context, as well as glosses and translations of all our examples.

Table 1. Type frequency of dative-subject predicates

Type Frequency

Old Norse-Icelandic


Archaic/Classical Latin


Ancient Greek


Old Russian


Old Lithuanian


The distribution of predicates across the five Indo-European languages is quite uneven, with Old Norse-Icelandic texts being extremely high in type frequency, and Old Lithuanian predicates being rather low in type frequency. Old Lithuanian texts go back to the 16th century, but they are only available to a limited extent in an electronic format. The Old Lithuanian part of the database is therefore quite small, substantially smaller than a corresponding database for Modern Lithuanian. We believe that there are more data to harvest not only from Old Lithuanian texts, but also from the other languages, although such data harvesting has not been possible at present. Our figures in Table 1 are therefore not to be regarded as exhaustive (although we treat them as such for the purpose of this methodological exercise).

Let us now consider the most common sememes, i.e. those that are found across three or more branches. These are listed below, first the ones documented in all five branches, then the ones found in four branches, and finally the ones found in three branches. The sememes found in two languages are too numerous to be listed.

Sememes found in five branches (ten types):

- like: falla (i geS) (ON-I), placere (Lat), areskein (Gr), patikti (OLith), ougazati (OR)

- be pleased: finnast (ON-I), libere (Lat), arestos einai (Gr), liuba buti (OLith), vbsladeti (OR)

- be proper for sby: sama (ON-I), decori esse (Lat), dikaion einai (Gr), tikti (OLith), gode byti (OR)

- be sufficient/suffice: duga (ON-I), satis esse (Lat), exarkein (Gr), kakti (OLith), dovblati (OR)

- be of shame: elenchos einai (Gr), vera skomm aS (ON-I), geda buti (OLith), ljute (OR), pudor esse (Lat)

- suit, become: sama (ON-I), commodus esse (Lat), prepein (Gr), dereti (OLith), prijetbnb, byti (OR)

- succeed: takast (ON-I), succedere (Lat), sumbainein (Gr), sektis (OLith), ougoditise (OR)

- lack: bila (ON-I), deesse (Lat), elleipein (Gr), stokoti (OLith), lixb byti (OR)

- need: vera porf a (ON-I), oportere (Lat), dein (Gr), reiketi (OLith), nadobe byti (OR)

- seem: virSast (ON-I), videri (Lat), dokein (Gr), regetis (OLith), javitise (OR) Sememes found in four branches (15 types):

- enjoy: philos einai (Gr), contingere (Lat), smagu buti (Lith), radostbnb (OR)

- find easy: facilis esse (Lat), oudobb byti (OR), anesis einai (Gr), lengva buti (OLith)

- be difficult: acerbus esse (Lat), argaleon einai (Gr), sunku buti (OLith), vbstouziti (OR)

- be in sorrow: oiktos empiptein (Gr), gailti (OLith), pecalb byti (OR), dolori esse (Lat)

- be in pain: algeinos einai (Gr), skaudeti (OLith), zblo byti (OR), dolor esse (Lat)

- be dear: carus esse (Lat), miela buti (OLith), Cbstbnb byti (OR), entimos einai (Gr)

- find (un)important for sby: skipta mali (ON-I), levis esse (Lat), luein (Gr), reikalinga buti (OLith)

- be of smb's business/concern: res esse (Lat), meteinai (Gr), rupeti (OLith), nadobbnyi byti (OR)

- benefit: aflast (ON-I), prodesse (Lat), kerdion einai (Gr), polbza byti (OR)

- have use of sth: verSa gagn aS (ON-I), lusitelein (Gr), naudinga buti (OLith), usus esse (Lat)

- occur to one's mind: hugsast (ON-I), succurrere (Lat), eperchesthai (Gr), pri-iti vb ourrrb (OR)

- know: vera (o)kunnleiki a (ON-I), (ig)notus esse (Lat), vedomo byti (OR), gnostos einai (Gr)

- happen: accedere (Lat), apobainein (Gr), atsitikti (OLith), loucitisja (OR)

- appear: birtast (ON-I), apparere (Lat), phainesthai (Gr), dozbreti (OR)

- have, possess: einai (Gr), buti (OLith), esse (Lat), byti (OR) Sememes found in three branches (14 types):

- dislike: leiSast (ON-I), displicere (Lat), nepatikti (OLith)

- have fear/agony: ofbjoSa (ON-I), timori esse (Lat), baisetis (OLith)

- be in danger: standa mein aS (ON-I), molestus esse (Lat), kindunos einai (Gr)

- be of good: bene esse (Lat), lusitelein (Gr), blago byti (OR)

- have problems with sth: vandraeSast (ON-I), laboriosus esse (Lat), metezbno (OR)

- find (im)possible: hlySast (ON-I), adunatein (Gr), mocbno byti (OR)

- find strength in sth: vera styrkur aS (ON-I), firmus esse (Lat), pora byti (OR)

- have a wish: optatus esse (Lat), poluaretos einai (Gr), xotetise (OR)

- get, receive: askotnast (ON-I), tekti (OLith), imati (OR)

- be allowed: licere (Lat), exeinai (Gr), lbze byti (OR)

- feel warm: hitna (ON-I), teplo byti (OR), therme einai (Gr)

- expect: vera von (ON-I), expectatio esse (Lat), epanamenein (Gr)

- be surprised: thauma einai (Gr), dyveti (OLith), divbnyi (OR)

- agree: lynda (ON-I), keisthai (Gr), constare (Lat)

The 232 sememes may be divided into the following 49 narrowly-defined lexical semantic verb classes (see Appendix), some of which can be subsumed under wider semantic category labels, given in boldface below:

Verbs denoting Emotions:

11. Verbs expressing burden/load


Verbs of liking/being pleased

12. Verbs of sorrow/sadness


Verbs of dislike

13. Verbs of pain


Verbs of longing

14. Verbs of bitterness/hate


Verbs of enjoyment/happiness

15. Verbs of shame


Verbs of feeling/experiencing

16. Verbs of care


Verbs expressing fear/danger

17. Verbs expressing hope/wish


Verbs denoting suffering/distress


Verbs expressing anger/irritation

Verbs of Gain:


Verbs of boredom/tiredness

18. Verbs of benefit


Verbs expressing relieve/ease

19. Verbs of permission

20. Verbs of growing

Verbs of Hindrance:

21. Verbs of hindrance

22. Verbs of failing

23. Verbs of slipping/losing

Verbs of Ontological States:

24. Verbs of (dis)similarity

25. Verbs expressing superiority

26. Verbs of properties/abilities

27. Verbs of other ontological states

Verbs of Evidentially:

28. Verbs of seeming/appearing

Verbs of Possession:

29. Verbs denoting possession

Verbs of Bodily States:

30. Verbs expressing bodily temperature

31. Verbs of getting better/worse (of illness)

32. Verbs of getting younger/older

33. Verbs of sleeping/being unconscious

34. Verbs of swallowing/choking

35. Verbs of symptoms of diseases

36. Verbs of hunger/thirst

Verbs of Cognition

37. Verbs of thinking/beginning to think

38. Verbs of (in)determinacy

39. Verbs of surprise/confusion

40. Verbs of knowing/change in knowledge

41. Verbs of agreeing/disagreeing

Verbs denoting Attitudes:

42. Verbs expressing sufficiency/usefulness

43. Verbs expressing appropriateness/ suitability

Verbs of Perception:

44. Verbs of perception

Verbs of Speaking:

45. Verbs of speaking

Verbs of Success:

46. Verbs of success:

Verbs of Happening:

47. Verbs of happening

Verbs of Modality:

48. Verbs of obligation

49. Verbs of lacking

Due to space limitations, we will not justify the classification given above, but refer the reader to Barðdal (2004) for details. There is no doubt, however, that a different type of classification may be applied to the material, equally successfully or unsuccessfully; at present we will let it suffice to say that for the purposes of comparison and reconstruction, the only thing of importance is that the same classification be applied consistently to all the languages being compared.

Before presenting an overview of how the various lexical semantic verb classes divide across the five early and archaic Indo-European language branches, let us first consider some different scenarios and what kinds of conclusions can be drawn from different types of semantic overlap.

Consider first Figures 1 and 2. If the Dative Subject Construction is not inherited, but an independent development in the various branches of Indo-European, then one would not necessarily expect any semantic overlap, as is shown in

Independent development: No overlap

Figure 1. Independent development: No overlap

Independent development: Overlap by chance: Lack of systematicity

Figure 2. Independent development: Overlap by chance: Lack of systematicity

Figure 1. Or, one would at least not expect more overlap than by pure chance, as is shown in Figure 2, where there is some overlap, although it is not systematic. Overlap by pure chance should be characterized by lack of systematicity.

Consider now Figures 3 and 4. If the Dative Subject Construction is a relatively recent common development, for instance from late Proto-Indo-European and relatively soon before the dispersal of the daughter languages, one might expect nearly a 100% overlap in semantic fields, as the vocabulary will not have had much time to develop. This scenario is found in Figure 3. However, on this scenario, one would also expect a high number of cognates across the branches, which is not the case here (cf. Bauer 2000, Luhr 2011), again speaking against this

Recent common development

Figure 3. Recent common development

Early inheritance: Productivity

Figure 4. Early inheritance: Productivity

Early Inheritance: Non-Productivity

Figure 5. Early Inheritance: Non-Productivity

Early Inheritance: Non-Productivity

Figure 6. Early Inheritance: Non-Productivity

scenario as being likely for the development of the Dative Subject Construction in Indo-European.[1]

Alternatively, if the Dative Subject Construction is an early inheritance, say from Early Proto-Indo-European, and not a recent development, one would expect a common semantic core, but with some branch-specific developments. This scenario, shown in Figure 4, presupposes that the Dative Subject Construction has been productive in the individual branches.

Moreover, if the Dative Subject Construction is an early inheritance from a common proto-stage, and has not been productive, there are two possible scenarios; first, one might expect either more or less the same semantic fields across the branches, provided that the relevant vocabulary has not fallen into disuse. This scenario is shown in Figure 5. Or, alternatively, if the vocabulary has fallen into disuse, one might expect the semantic space to have shrunk in some of the branches. This last scenario might look like the one presented in Figure 6.

Finally, the last scenario to be discussed here involves contact and mutual borrowing. We do of course know that there has been contact among the different Indo-European languages, as between Ancient Greek and Latin, on the one hand, and Germanic, Baltic, and Slavic, on the other. However, a contact situation that entails mutual borrowing should result in a number of cognates across the languages in contact, and their form should be easily recognizable as borrowed. The same is true for areal contact; it should also result in identifiable cognates across languages geographically adjacent to each other. Such a situation is not at hand here. If anything, the lack of cognates across the early Indo-European languages has been the subject of some discussion in the literature (cf. Bauer 2000, Luhr 2011), although a few cognates are now in the process of being documented. These cognates are not borrowed but show clear signs of rule-based language-dependent phonological developments (Barðdal & Smitherman 2013) and must therefore be regarded as being old in these languages.

So what kind of semantic overlap is found when we compare the lexical semantic verb classes found with the Dative Subject Construction across the five Indo-European branches under investigation here? What does the picture revealed by our five branches look like? This is the topic of the next section.

  • [1] In contrast to the claims of Bauer (2000) and Luhr (2011) who argue that cognates across the early IE languages are almost nonexistent, BarSdal & Smitherman (2013) have in fact documented a few well-defined sets of cognate predicates instantiating the Dative Subject Construction across the early Indo-European languages. To be sure, the number of cognates is not high, approximately a dozen or so, at this initial stage of their research
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
Business & Finance
Computer Science
Language & Literature
Political science