Old Frisian Compounds: Kinship Terms

Together with derivation, compounding is one of the most important word-building processes in the world languages. In the Germanic languages, compounding has become extremely productive (Müller, 2015). As opposed to studies in other Old Germanic languages, compounding in Old Frisian (OFr) has not been investigated in full. Unlike in the neighbouring languages, only a small number of OFr lexemes is present in the most comprehensive book on Old Germanic nominal compounds (Carr, 1939). The semantic field of kinship terms includes basic concepts reflected in the lexis of a language. A lot of OFr kinship terms are compounds. The questions posed in this paper are what compounding models are used to build up OFr kinship terms: (1) in terms of the Sanskrit classification, (2) in terms of parts of speech used as the constituents, (3) in terms of the semantic connection between the constituents. In order to do that, compound structures are split into constituents and then formally analyzed. Statistical analysis reflects the productivity of compounding patterns and the balance between the trends. Thereafter, the formal data are compared to the semantic patterns identified within the compounds. The end goal is to assess the degree of the interconnection between the morphological and semantic levels. The data can be used to compare the trends in OFr and in the related Old Germanic languages in order to contribute to the theory of compounding in Germanic.


Introduction
West, East and North Frisian are Germanic languages nowadays spoken in the Netherlands and in Germany.Importantly, West Frisian has an official status in the Netherlands, and it is a minority language in focus of the Frisian Academy (Fryske Akademy) in Leeuwarden.
Having been a language surrounded by larger Germanic languages, the historical forms of Frisian have not been paid as much attention, as, e. g., those of English or Dutch.Yet Old Frisian is the closest relative of Old English, preserving the early features of the Anglo-Frisian language cluster.By now, Frisian and English have been under the influence of Dutch and French, respectively, for centuries.
The data on Old Frisian are specific for several reasons.First of all, it is unique in its periodization, namely OFr is roughly dated between 1250 and 1550, when other Germanic languages were at their Middle stage of development.Second, OFr as a whole is not as well presented in Germanic studies as its relatives.Thus, a research is needed to fill the gap in the description of the Old Germanic languages.Third, the texts written in OFr belong to the domain of law.
Although there have been research projects conducted on such data, the works have been dedicated mostly to legal features, etymology, lingo-cultural peculiarities, anthropological aspects or to usage of a certain word / word combination.The OFr morphological system merits a comprehensive study, which allows linguists to perceive Old Frisian word-building processes in a transparent way, just like those in other well-studied Old Germanic languages -Gothic, Old English, Old High German, etc.

Old Frisian Word Stock
According to the database compiled on the basis of the dictionary of Hofmann / Popkema (Hofmann, Popkema, 2008), compound words amount for 36 % of the total Old Frisian word stock.Furthermore, the major part of them is nominal compounds, 63 %.To be more precise, nominal compounds are roughly 23 % of the total OFr word stock.
It is understandable, as nouns in general denote notions and objects.Interestingly, 44 % of all the nouns are compounds which is almost half of the nominal word stock.It means that roughly every even nominal notion is rendered with a compound word.
The lemmas presented in this article are normalized in terms of spelling, based on Hofmann / Popkema's dictionary (Hofmann & Popkema, 2008).
The overall aim of the research as a larger project is to describe the compounding patterns in OFr, as per the 3 classifications widely used in Indo-European studies and language typology: (1) the Sanskrit classification, (2) classification in terms of parts of speech used as the constituents, (3) classification in terms of the semantic connection between the constituents.As a result, pattern productivity will be statistically reflected.Next, remarks on the interconnection between the semantics and structure of compounds will be made.Thus, the form of OFr compounds will be treated not only in a descriptive way, but also in terms of the morpho-semantic relation.
In a more general words, the main question to pose is "How is the choice of a certain compounding pattern made in order to render a notion?"and, if put in a more practical way, "Are there any semantic peculiarities typical of a certain compounding pattern?" By answering the questions in the domain of OFr, not only a gap in morphology (compounding theory) of the Old Germanic languages will be filled, but also a more detailed picture will be available for the theory of compounding in the World's languages.

Compounding Theory: Defining a (Compound) Word
When dealing with compound words we inevitably face the issue of differentiating between the so-called 'graphical word' and the 'word' as a notional unity.This can be explained by compound words being built by stem or root composition of two or more words.As a result, they are not easily differentiated from the word combinations consisting of the same components in the full and agreed form.In terms of diachronic development, the difference between word combinations and compounds is of a specific interest, as "the transition from syntactical juxtaposition to true composition is very gradual; no sharp line of demarcation can be drawn between them.This is proved by the great uncertainty which prevails in the orthography of modern languages with regard to the treatment of various combinations as compounds or as separate words.This uncertainty has led to an orthographical compromise -the use of the hyphen."1Quoted above, H. Paul also emphasizes falseness of the impression that elements of a compound word are pronounced in a more continuous way compared to words in a word combination.This phonetic feature can be explained by the stress being on the main component, which can be seen in the combination of "a link word + a notional word" forming one cola, i. e. we can see a trend somehow similar to Wackernagel's law applicable to the Proto-Indo-European language (clitics appear in second position to the first syntactic phrase or stressed word).The stress can fall on any component of a compound word, the only category being excepted is compounds formed by analogy with an already existing model.Thus, "we have to seek for the distinction between a compound and a group of words united under one main accent, not in any physiological cause, but in the conditions of the psychological grouping.The only essential point is that the whole as such being in some way isolated from the elements of which it is composed.The degree of isolation necessary in order to cause the fusion to pass into a compound cannot be expressed in any universally applicable definition".2(Paul, 1891, 372-373).
There have been discussions on the criteria for differentiating between word phrases and compounds; yet, there is no final solution for this definition problem (Lieber, Štekauer, 2009, 14).
Leaving this matter aside for further discussions, we should pay attention to specific Germanic features that allow compounding to become productive as a model.
It is crucial that the stress in the Old Germanic languages was fixed and fell mostly on the first (root) syllable.Due to this phonetic pattern the post-tonic part of the word was uttered in a less clear way, leading to full or partitive suffix and ending reduction.Thus, there was a gradual change of the general phonetic / phonological layout of the word, as a result of which the old word building system with suffixation as the main tool was becoming less and less productive.These conditions were favourable for compounding to become more productive.
Another important matter is differentiating between compound words and noncompounds/word phrases.As both the semantic and grammatical levels of research are taken into consideration here, it is necessary to note there are formal criteria for differentiating between compounds and word combinations in terms of word building: as has already been mentioned, the current research does not focus on the disjoint spelling of lexical units, which is formalistic, but about phonetic (stress in the cola / compound words) and morphological features (element transformation at the junction of compound elements, various mutual influences).
For example, in order to analyze the opposition of word combinations and compounds, the following semantic group can be taken a look at: in the denotation of one of OFr oaths, the notion is rendered with the word combination [adjective + noun], where each word is likely to be stressed.The reason is that the adjective does not just make the meaning more specific, but is the actual unit denoting what kind ISSN 1392-8600 E-ISSN 1822-7805 Žmogus ir žodis / Svetimosios kalbos / 2017, t. 19, Nr. 3 of oath it is -fremethe ēth 'oath of not being affined' (fremethe 'fremd; anverwandt', 'someone else's; not affined').In case the unit has transformed into a compound, the main stress is likely to fall on the main component (like in Modern German) or on the first constituent (like in Modern Frisian), as the whole term is perceived as one unityfulēth 'full oath, oath of 12' (fel(o), ful 'viel', 'many').No morphological changes (case ending reduction, insertion of a connecting vowel, etc.) can be seen at the junction of the elements.
When the interconnection between the elements becomes psychologically closer, they actually start being perceived as an inseparable unit, especially if one component is no longer used as a notional word in the course of the language development over time, and these are cases of grammaticalization, e. g., Modern English -dom at a certain point of time turned into a productive suffix.
The question is: what is the reason for closely related notions to be denoted by different morpho-syntactic models?The question arises when dealing with almost any Old Frisian term category.Consider the following example.Popkema (2007) provides a detailed description of a legal category, which mainly contains oath terms.Most of them (27 out of 34) are compounds built with -ēth / -ēd (dialectal variation) 'oath' (Eid) as the second element, or word combinations with a more or less equivalent morphosyntactic connection between the components: compare ēth der wird 'oath of witness' and rēdsliōdenaēth 'oath of councilors'.
Fogedēd 'bailiff's oath' (Vogteid) is also derivationally and semantically linked to dēd 'admission (for example, of offence)' (Tateid).Thus, the main word building model for such terms is [... + 'oath'] composition, where in almost all the cases the first component is a noun, which denotes a person giving an oath or a relevant judicial case.
An exception to the [... + 'oath'] model is wīthe 'relics; oath on relics' (Eid auf die Reliquien), still with the equivalent wīthēth (Holthausen, 1925, 131).Here we can see an example of metonymy (i.e. meaning transfer from a compound to its part used as a simplex) typical of the Old Germanic languages which, nevertheless, breaks the symmetry of the morphological system: wīthe is a derivate from wīa 'make holy', not from a substantive (Popkema, 2007, 353).The same is said about lāda (lēde) 'management; oath of fair management of the entrusted property' from the verb lēda 'manage'.
Here are three more examples demonstrating metonymy.Stef means 'an alliterative oath'.The word itself means 'stick, pole; oath' (Holthausen, 1925, 103) and occurs in et steve stonda 'give an oath (liter.'stand by an oath')' (Popkema, 2007, 352).The second terms is handtr(ī)owe 'oath with touching a hand of a kin's member', the second component, tr(ī)owe, meaning 'agreement; loyalty' (Holthausen, 1925, 116).Finally, the third term is kest 'choice; facultative oath'.Here, we can see metaphoric transfers of meaning, needless to mention polysemous riucht, among others meaning 'purification oath'.As we can see, the principles of analogy (for instance, the same second component in compounds) and association (for example, transfer of meaning) result in the interdependence of the semantic and morpho-syntactic levels of Old Frisian.Another important question is what interconnection is between the parts of speech of the first word in the word combination, as well as how likely it is to fuse with the second component.

Types of Compounds and their Origin
The analysis of language data lies at the edge of two aspects, diachronic and synchronic.Saying 'diachronic', we mean the sum of all semantic-grammatical changes occurring within the development from the Proto-Germanic language to (Old) Frisian in its variants; saying 'synchronic', we mean contemporary interconnections in the Old Frisian language system (circa 1150-1550 AD) and within the West Germanic language community as opposed to the related languages at the comparable stages of their development.
Based on this, we should start with the description of the compound types existing in early Germanic languages and give a more general interpretation of the abovementioned scheme: 1. copulatives, substantive + substantive; 2. determinatives: (a) substantive + substantive, (b) adjective + substantive, (c) substantive + adjective, (d) adjective + adjective; 3. exocentrics, adjective bahuvrîhis, with the first component being a numeral, adjective or substantive (Carr, 1939, 161).In the same period, a new model started developing in West Germanic languages, which is [verb + substantive].One of the types, its nature and existence at the early stage of the Germanic language development, substantive bahuvrîhis, is analyzed in detail in the comprehensive book Nominal compounds in Germanic by Ch. T. Carr (1939).As convincingly shown by the author, the origin of this type of compound words cannot be explained by the change of the determinative model [adjective + substantive] (mutated compounds), bahuvrîhis appear in the languages at the same time as determinative compounds.One more important point is that not all bahuvrîhis have equivalents among determinative compounds.In the West Germanic region, the [adjective + substantive] type is still remaining and developing, whereas the use of the adjective bahuvrîhis is fading away (Carr, 1939, 163).Another important observation is that "the nominal compounds with an adjective as the first member were but few in number in early Germanic and have increased enormously in modern times" (Carr, 1939, 163).In this paper, the units will be analyzed in terms of the composition according to the part of speech.
Initially, the substantive bahuvrîhi model was used as the method of name-giving, which is described in detail by T. V. Toporova.Among related morpho-syntactic features we can mention adding compounds to a proper name (Ger.Harald Schönhaar), and change of accentuation as a result of the notional stress moving from one component to an/ the other.In the beginning, the function of bahuvrîhi had to do with describing the 'mystery qualities, character' (heimlicher Adjectivcharacter) of a certain person.It led to a difference between the declination type of such determinative compounds and the noun declination pattern of the second compound (in case of a simple word).The adjective bahuvrîhis was modified by the addition of various determinative suffixes (-ja, -jô, -ig, -isk, -līc), endings of past participles (-ed, -od), or were substantivized by adding the ending -an.However, in Carr's opinion, namely based on his explanation of the origin of the bahuvrîhi type, there is no actual difference between the substantive and determinative bahuvrîhi�3 Bahuvrîhis were mostly nicknames describing a person's peculiarities, i. e. they belonged to the spoken language of a lower style (Scandinavian nicknames and West Germanic kennings should be discussed separately), which can explain why the number of such compounds recorded in written sources is so small. 4he so-called 'imperative compounds' can structurally be classified as the [verb + substantive] model.They were barely recorded either, the reason being their 'jocular and abusive' meaning.

Components of Compounds: Parts of Speech
Furthermore, compounding typology refers to the parts of speech that constitute a structure.These parts of speech can be substantives, adjectives and verbs in the following combinations (starting with the most frequent model): 1� substantive + substantive; 2� adjective + substantive; 3� verb + noun.
There is one more, a slightly different (compared to the above-mentioned classification of J. Grimm  , 1963, 115-116).Whereas the first model remains in the West Germanic languages only rudimentary, the second and the third ones are productive to a roughly equal extent.
In the [verb + noun] model, the first component is a deverbative which makes such compounds typologically close to the [substantive + substantive] model."As clearly brought out in Ch.Carr's monography, the model with a verb in the first part of the composition, not being common Germanic, can be seen as an innovation of the West Germanic languages". 5As mentioned before, the existence or non-existence of a connecting vowel can be explained in several ways, so when differentiating between the compounds with a substantive / verb as any of the components it is possible to rely on the word semantics (e. g., the compounds with the second element meaning 'a stroke', can be classified both as [... + noun] and [... + verb]).
The [adjective + noun] model, although found in all Old Germanic languages, is far behind the first of the two letter groups in terms of frequency of use.As for West Germanic, it is statistically more typical to stick to a composition with a pure root as the first component, although we can see pattern inconsistencies documented in Old Frisian (ald(e)-feder 'Grossvater', 'grandfather').It is an often dilemma to identify the model as some word forms can fall together (within the paradigm, in parts of speech, etc.).
It is quite difficult to classify the types of relations between the parts of a compound noun in Old Frisian (see Carr, 1939, 320), that is why we will skip this classification here.

Components of Compounds: Semantics
First of all, we must emphasize how vague the borderline between 'a semantic, syntactic unity' and 'a compound' is.Carr suggests that to differentiate between them the following criterion can be used: the meaning of a compound is not equal to the sum of meanings of its components which sometimes leads to semantic polysemy and, as a result, lack of clarity.6 Besides, in some cases the meaning of a compound can be explained only through a long and cumbersome construction [ibid.]; it is even more striking when translating terms into Russian, where other means are used to render such notions.
In general, all compounds can be classified into 4 groups: In relation to compounding, grammaticalization plays a major role: one of the components in models (ii), (iii) and (iv) can fade in its independence and develop into an intensifying prefix or suffix.8 The most remarkable example is the development of lвs 'free' into a suffix meaning a lack of a quality vs. līk 'image, form' turning into a suffix as opposite meaning something similar, or with a quality present.Whether we consider these words to be suffixes or notional words is crucial for the analysis, as it has to do with the conceptual decision on what is considered a compound and what is not.

Germanic Compounds: Carr's Classification
Although significantly updated since 1939 for the world languages in general, the Old Indian classification of compounds still remains relevant (Lieber, Štekauer, 2009, 331-332), as well as is the classification specified for Old Germanic languages by Ch. T. Carr (1939).The reason is that not all the Indo-European compounding models have been developing in Germanic languages, so the amendments done for the universal compounding classification do not have any effect on the limited number of the compounding models found in Germanic.
A large number of Old Frisian compounds can be united into several semantic groups, the basic lexicological ones of which are: kinship terms, representatives of authority, parts of the body and specific legal terms (common law, law of crimes).
Dealing with Indo-European lexis, it is reasonable to begin with kinship terms as the oldest terms present in all Indo-European languages.One of the goals of this article is to classify Old Frisian compound kinship terms in line with Carr's classification.
The description of Proto-Germanic words reflecting kinship can be found in The Proto-Germanic Words Inherited from (sic!) Protoindo-European Which Refl ect the Social and Economic Status of the Speakers by W. P. Lehmann; yet, neither Old Frisian terms nor compounds are present there.
Not to go into the question at large, we have marked case endings / connecting vowels in the terms presented in the tables.A research dedicated to the issue could be pursued on a vast scale.
According to anthropological linguistics, it is important to distinguish between real kinship and relationship by marriage (Lehmann, 1968, 16).The importance stems from the system in line with which an heir entered into his estate.In this article, mostly real kinship terms will be considered.
The terms connected with the notions 'father' and 'mother' are listed in Table 1.As we can see, words like fadera 'godfather' can have a more general meaning 'father' when used as a component of a compound term.* not a compound, given for comparison with other naming units for 'nephew'.
The status of a newborn determined their life: there is a gradation from êgene-bern 'unfree child', to hōrkind 'bastard' set against Germanic jonchêre 'young Master of noble birth'.In terms of the overall results gained in the three tables, we can see that differentiating between Karmadhâraya and Tatpuruša indeed is a challenging task.Yet tracking the development of these models can shed light on such an important language process as case ending of the first constituent developing into a linking element.For instance, in lemmas like mōderesbrōther 'mother's brother', -es becomes a case ending, it does not appear as a flexion of a feminine noun based on the paradigm.This might indicate that the element is not a case ending anymore, and rather acts as a linking element with the Genitive meaning (Lieber, Štekauer, 2009, 331).
Another important conclusion is that the linking element -es is frequently inconsistent in the first component: both forms with and without it are present in the texts, as reflected by the dictionary.
Next, a semantic connection between the constituents should be considered.As can be seen in the tables, most compound words in this semantic field belong to the [noun + noun] model.Although flexions / linking elements between the components in some cases can be either documented or not, the meaning of the lemmas indicates that the connection between the constituents is a determinative one.We assume it should have to do with the Genitive meaning of such compounds as 'X of Y', when 'Y' is well-known and definite in the context.Thus, the meaning of the first constituent is not attributive (=describing the qualities or availability of certain objects of the person); yet it equals the meaning of a word combination.For instance, brōthereskind is 'a / the child of the brother', with the first constituent being determinative, whereas knapakind is appositional, 'both a child and a boy, i. e. a male child'.
With reference to the above-mentioned, it seems reasonable to highlight the link between the meaning of a compound and its semantic type.As has already been mentioned, the linking elements are not necessarily present in determinative compounds; nevertheless, the inconsistency of the representation of the linking elements may indicate how Tatpuruša and Karmadhâraya were blurred over time.The determinative compounds, nevertheless, found their way into Modern Frisian (Hoekstra, 2003) and consistently retain -s in between the constituents as a linking element.

Conclusions
The aim of research on compounding in Old Frisian, presented in this article, is to describe the compounding patterns in Old Frisian and link them to the semantics of the lemmas.75 Old Frisian compound lemmas from the semantic field of kinship terms have been analyzed.
The method of our analysis is to split the compounds into constituents and classify them in a formal way.All the analyzed lemmas can be split into 3 groups based on their structure: ISSN 1392-8600 E-ISSN 1822-7805 Žmogus ir žodis / Svetimosios kalbos / 2017, t. 19, Nr. 3 / Svetimosios kalbos / 2017, t. 19, Nr. 3 ) grouping system of substantive compounds: a) the model with the first component being a noun stem, b) the model with the first component being a noun root, c) the model with the first component being one of the objective cases of a noun (СГГЯ / SGGJA / Svetimosios kalbos / 2017, t. 19, Nr. 3
with the co-ordinate interconnection, the same as Dvandva in Sanskrit;(ii) with the first component performing a limiting function and linked to the second component by the case or prepositional connection, i. e. Tatpuruša.The types (iv) with the components functioning as apposition to each other, each of them denoting shades of meaning of the same notion, genus proximum included, as well as pleonastic compounds with the elements being similar in their meaning.Over time, this type can fuse into tautological compound words with synonymic components.The models are: [substantive + substantive], [adjective + adjective], with the second component being compared to the first one (Old High German grasegruoni 'green as grass').

Table 1 .
Terms associated with the notions 'father' and 'mother'

Table 3 .
Terms associated with the notion 'child'