System-functional stratification of the average Turkish-Russian dictionary

DOI:

The lexicon of the language is measured in tens or even hundreds of thousands words and seems to be unusable for comparative lexicology. Therefore, one of the important tasks of modern lexicology is to create well-ordered descriptions of vocabulary, that enable distinguishing its representative cores of about 1000 words, providing such a comparison. And if there is no shortage of dictionaries containing information about the vocabulary of the world’s languages, the necessary tools for theoretical mastering of this information – Parametric Analysis of Lexicon (hereinafter – PAL) – appeared relatively recently [Titov, 2002; 2004a]. One example of the use of this toolkit is the collective monograph [Kretov et al., 2016].

Parametric analysis of the Turkish vocabulary is presented in a number of papers [Bugaev, 2006; Kretov et al., 2016; Semenova, 2018]. However, in these studies, the object of analysis were Small Turkish-Russian dictionaries measuring about 10,000 words: Turkish-Russian and Russian-Turkish Dictionary (Rybalchenko, 2001) was investigated by V.P. Bugaev [Bugaev, 2006] and by I.D. Semenova [Semenova, 2018], Brief Turkish-Russian Dictionary (Scherbinin, 1977) was investigated in the collective monograph [Kretov et al., 2016, p. 411].

In this regard, it seems appropriate to explore a larger Turkish-Russian dictionary and put the parametric analysis of Turkish vocabulary on a more complete and more modern basis. The purpose of this article is to study the connections that make up the lexicon system of the Turkish language and to stratify the vocabulary of the source dictionary according to the systemic weight of the components of its words.

Data and methods

The object of the study is the “Turkish-Russian Dictionary” (Yusipova, 2005), rich by information and the most modern of the available Turkish-Russian dictionaries of this type. When counting one-word lemmas (without lemmas-phrases and reference articles), the volume of the dictionary has 25,097 words. The choice of bilingual dictionary is conditioned by the need for a single basis to compare the lexicons of Turkic languages, both among themselves and with the lexicons of any other languages of the world. For this purpose the Russian language, which performs the function of meta-language, has been accepted.

We proceed from the widespread notion of a field organization of the world’s languages vocabulary the principle of “core in cores”, that is according to the fractal principle. The core of the vocabulary of any language is the root words, followed by the derivative words sector, and finally, the periphery of the lexical-semantic system is formed by composite nominations (including phrases). Peripheral vocabulary can change as quickly as possible, but it does not affect the core of the lexical-semantic system, the selection of which is the purpose of our analysis. In comparative historical linguistics the change of the lexical core of language (“basic vocabulary”) is recognized as the most important event that can occur in language and with language: “Such cases are known and invariably qualified as a change of language. <...> ...If the basic vocabulary begins to be actively borrowed, the rest of the vocabulary of the language tends to be saturated with borrowings even more... as a result there is virtually nothing left of the original language – it can be stated that the people have switched to another language” (here and further English translation is ours. – Z. D., A. K.) [Burlak, Starostin, 2005, p. 14].

The subject of the study is the system-forming parameters of Turkish vocabulary. The method of research is Parametric Analysis of Lexicon (PAL), described, substantiated and tested in studies [Titov, 2002; 2004a; Voevudskaya, 2015; Kretov, 2011; 2017; Kretov, Cherechecha, 2020; Kretov et al., 2016; Kretov, Gasuns, Leonchenko, 2021; Merkulova, 2018; Semenova, 2018]. PAL is a method of analyzing vocabulary according to the data of foreign-Russian dictionaries. As part of the parametric analysis of vocabulary, the indicators of different dictionaries of the same language were repeatedly compared in order to assess the ratio of objectivity and subjectivity of their data. The result of the research is: “lexicographic sources, on average, by 2/3 reflect the realities of the language’s lexical system, and only 1/3 of the information they contain depends on the subjective factor” [Voevudskaya, 2015, p. 206].

Thus, all existing bilingual dictionaries representing a subjective image of objective reality today are the only source of information for the construction of lexical-semantic typology of languages. The dictionaries have already analyzed and represented the epidigmatics (polysemy) and (although sparingly) syntagmatics (and in implicit form – also paradigmatics) of the vocabulary of the corpus of texts that formed the basis of the dictionary file. As a parametric analysis of foreign-Russian bilingual dictionaries, PAL accepts each of the dictionaries analyzed, including (Yusipova, 2005), and criticism of source dictionaries is carried out post factum – through comparison with the results of analysis of other dictionaries (see: [Titov, 2004b]). PAL assumes the definition of four private parametric scales for each word represented in the dictionary by its vocabulary form – lemma: functional weight (indirectly estimated by the length of the lemma: the shorter is the lemma, the greater is F-weight), paradigmatic weight (P-weight is indirectly estimated by the number of synonyms for a given lemma), syntagmatic weight (S-weight is indirectly estimated by the number of combinations with this lemma in a dictionary article, including illustrative examples) and epidigmatic weight (E-weight is indirectly estimated by the number of meanings allocated by lemma in the dictionary article). The addition of private weights of each lemma gives integral parametric weight (I-weight).

Each of the scales is calculated on the same formula:

Pr_i =

Σ r - R 1 - i

Σ r

where Σ r – the sum of lemmas of all ranks, R _{1– i} – the sum of lemmas from the first rank to the given, and Pr_i – the weight of the lemmas of the i -rank. Pr_i values fluctuate in the interval from 0 to 1.

The logic of the formula is simple: the fewer participants showed the same or better result, the higher the place (rank) of the participant. The weight of the lemmas of each rank depends on the number and weights of the lemmas of all other ranks. Thus, each of the words (lemmas) in the dictionary affects the weight of all the other words (lemmas) for each of the 4 parameters. This approach sharply narrows the freedom of research arbitrariness, increasing the scientific objectivity of the study.

Results and discussion

This section is devoted to description of the analysis results for each parameter of the source dictionary, consisting in “weighing” each word within this particular parameter. Syntagmatic and paradigmatic connections are system-forming for the dictionary in synchrony, mutually defining each other: syntagmatics is represented by speech sequences, and paradigmatics is represented by synonymous and other sets of words similar in any respect. Epidigmatic (derivational in a broad sense) connections characterize the dictionary as a developing and self-expanding object, which is associated with diachrony, and the functional parameter characterizes the dictionary as the most important part of a living, i.e. functioning language in the process of communication. Thus, a set of four system-forming parameters characterizes the dictionary as a developing, self-expanding and functioning system. Their totality provides the possibility of “weighing” words by their system-forming (integral) weight, i.e. by their place and importance in the lexico-semantic system. The novelty and scientific value of the results are presented at the end of each section.

Function stratification of Turkish vocabulary

The word usage is an unobservable factor. The frequency of the word in any text is equivalent to two independent patterns: objective – linguistic and subjective – textual. The author of the text has power only over subjective regularity. The objective one is imposed by the language: in any Russian text the most common word will be i ‘and’, in any English – the, in any Turkish – bir ‘one, some’ (Göz, 2003). This information does not give anything to highlight the lexical cores of the language. That is why in parametric analysis of vocabulary it is more expedient to evaluate the usage of words on such an objective observed parameter as the length of the lemma

(representing the word in the dictionary): over the length of the full-digit word (as opposed to its frequency) the author of the text is not in power.

Functional stratification of vocabulary raises the question: in which units to determine the length of the lemma. For the Turkish language this question can be removed: adopted in 1928 latinized Turkish alphabet quite accurately reflects the sound composition of Turkish speech. In the mass (many thousands!) study of the Turkish vocabulary, we have the right to put an equal sign between the length of the Turkish lemma in letters and its length in sounds. The prospect of using these parametric analysis of the vocabulary of the source dictionary in comparison with other Turkic languages, especially the Kyrgyz language, forces us to deviate from the form of lemmas in the dictionarysource to ensure that the result is comparable to the dictionaries of those Turkic languages in which the verb form is given in its purest form and marked with a hyphen. When calculating the length of the lemmas and calculating their functional weight (F-weight), -mak / -mek morphemes did not affect the length of the verb lemmas and their F-weight: the length of each lemma with these affixes was reduced by 3 letters of sound. For example, the length of the lemma çıkarmak ‘pull out, take out, extract smth.’ is not 8, but 5, the length of the lemma tutmak ‘hold’ – not 6, but 3, the length of the lemma almak (‘take’) is not 5, but 2 letters. The distribution of lemmas in the dictionary-source by length in letters based on the functional transformation supposition is presented in Table 1.

Table 1. Distribution of the source dictionary lemmas by length

Letters	Lemme	Cumul.	F-weight	Example	Meaning
1	5	5	0,99980	o	he, she, it
2	161	166	0,99339	aç	hungry
3	934	1100	0,95617	ana	mother
4	1838	2938	0,88293	açık	outdoor
5	4703	7641	0,69554	abıru	honor, dignity
6	4258	11899	0,52588	açınım	development
7	4417	16316	0,34988	adamlık	humanity
8	3863	20179	0,19596	adaletli	just
9	2153	22332	0,11017	akrabalık	kinship
10	1361	23693	0,05594	alışkanlık	habit
11	794	24487	0,02431	adaletlilik	justice
12	317	24804	0,01167	bağdaştırıcı	adapter
13	171	24975	0,00486	cesaretsizlik	indecision, timidity
14	66	25041	0,00223	dayanışmacılık	solidarity
15	38	25079	0,00072	değerlendirilme	score
16	14	25093	0,00016	rutubetlendirici	humidifier
17	2	25095	0,00008	toplumsallaştırma	nationalization
18	1	25096	0,00004	elektrokardiyogram	electrocardiogamma
21	1	25097	0,00000	erkânıharbiyeiumumiye	General Staff

The distribution of lemmas by length in the source dictionary is presented in Figure 1.

Figure 1 indicates the heterogeneity of word distribution in the dictionary, as evidenced by the presence of two peaks: 5 (mode) and 7. Since the words of spoken speech are frequent and therefore short, the formation of the second peak of the distribution with a length of 7 indicates the prevalence of longer derived words characteristic of written speech.

The heterogeneity of the distribution of words by length also indicates the genetic heterogeneity of the vocabulary, the formation of the vocabulary of the standard language is largely due to derivative and borrowed words: “Borrowed words in the Turkish language are represented mainly by Arabic and Persian vocabulary, the number of which in the 17th and 19th centuries reached 80–90% in some works. <...> The oldest lexical borrowings from European languages are acquisitions from the Greek language... <...> Borrowings from Armenian, Albanian, Hungarian, Romanian, South Slavic and Russian languages played a role in the formation of the dictionary of modern standard Turkish” [Kononov, 1997, pp. 409-410].

The shortest (that means – the most important, having the biggest F-weight) Turkish content words are two-letter: aç ( mak ) ‘open’; aç ‘hungry’; ad ‘name’; af ‘forgiveness’; ağ ‘net’; ağ ( mak ) ‘rise up’; ak ( mak ) ‘flow, pour’; ak ‘white’; al ( mak ) ‘take’; al ‘scarlet’; al ‘cunning’; an ‘moment’; an ‘reason’; an ( mak ) ‘remember someone’; ar ‘shame, modesty’; as ( mak ) ‘hang’; as ‘ermine’; aş ( mak ) ‘overcome’; aş ‘food; at ( mak ) ‘throw’; at

‘horse’; av ‘hunting’; ay ( mak ) ‘regain consciousness’; ay ‘moon’; az ( mak ) ‘become violent’; az ‘insufficient’; de ( mek ) ‘talk, say’; eğ ( mek ) ‘tilt’; ek ‘supplement’; ek ( mek ) ‘sow’; el ‘hand(s)’; el ‘stranger’; em ( mek ) ‘suck’; em ‘medicinal remedy’; en ‘width’ en ‘brand for cattle’; er ‘man’; er ( mek ) ‘reach sth’; es ( mek ) ‘blow (about the wind)’ ; eş ‘couple, partner’; eş ( mek ) ‘rake the ground’ ; eş ( mek ) ‘gallop’; et ( mek ) ‘do’; et ‘meat’; ev ‘house’; ev ( mek ) ‘hurry’; ez ( mek ) ‘crush, mash’; iç ‘the inside, the inside (of something)’; iç ( mek ) ‘drink’; iğ ‘spindle’; ıh ( mak ) ‘kneel (about a camel)’; il ( mek ) ‘weakly tie’; il ‘(administrative unit in Turkey) il ‘vilayet, province’; il ( mek ) ‘weak knot’; im ‘sign, signal’; in ( mek ) ‘go down’; in ‘den, hole’; ip ‘rope’; is ‘soot’; iş ‘work, labor’; it ‘dog’; it ( mek ) ‘push’; iv ( mek ) ‘hurry’; iz ‘trace’; öç ‘revenge’; od ‘fire’; öd ‘bile’; öd ‘smell of the burning the scarlet tree’; ok ‘arrow’; ol ( mak ) ‘to be, to happen’; öl ( mek ) ‘to die’; öl ‘soil moisture’; om ‘thickened/rounded end of the bone’; on ( mak ) ‘improve, correct’; ön ‘place (in front of something)’; öp ( mek ) ‘kiss ‘whom’; ör ( mek ) ‘knit’; ot ‘grass’; öt ( mek ) ‘sing; chirp’; ov ( mak ) ‘knead; rub’; öv ( mek ) ‘praise’; oy ‘opinion’; oy ( mak ) ‘make a’ recess/deepening’; öz ‘the essence (of a person)’; öz ‘native (about relatives)’; öz ‘river, stream’; sı ( mak ) ‘smash, break’; su ‘water’; ti ‘bugle signal’ ; uç ( mak ) ‘fly’; uç ‘the point, the pointed end (of a knife, etc.)’; um ( mak ) ‘hope, hope for someone’; un ( mak ) ‘organize’; un ‘flour’; ün ‘voice, sound’; ün ’ fame’; ur ‘neoplasm, tumor’; us ‘mind’; üs ’ base’; üş ( mek ) ‘to gather in a crowd’;

Length of lemmas in letters

Fig. 1. Distribution of lemmata by length in the dictionary-source

ut ‘shame’; ut ‘ud’ ; ut ( mak ) ‘to win’; üt ( mek ) ‘to scorch, burn with flame’; üt ( mek ) ‘to win in the game’; uy ( mak ) ‘to match’; uz ‘good; beautiful’; üz ( mek ) ‘to upset’; ye ( mek ) ‘to eat’; yu ( mak ) ‘wash’. It follows from the table that the functional core of vocabulary in the dictionary source consists of words that are no longer than 3 letters long. After excluding the vocabulary groups described above from this set, the size of the F-core was 983 words.

Syntagmatic stratification of Turkish vocabulary

Usually even explanatory dictionary stingily reflects the word compatibility. However, the degree of completeness-wealth of the representation of the syntagmatics of dictionaries does not affect the objectivity of their data: the most syntagmatically rich words remain so, no matter how many of their phraseological combinations (hereinafter PhC) are taken into account: 100 or 10. The less syntagmatically important words have no phraseological combinations. Another thing is that those words that, with a maximum of 10 PhC, had 0 phrases, in a dictionary with a maximum of 100 can have from 1 to 9 phrases. The scale and details of the syntagmatic curve change depending on the completeness of the data, but the form of the curve (in its objective part) remains the same: this is the idea of parametric “weighting” of words, and this is the objectivity of the data obtained at such weighting. The limitation of syntagmatic information in bilingual dictionaries makes the syntagmatic “weighing” of words take into account all the vocabulary evidence of compatibility presented in the dictionary: both stable phrases with the word, and the compatibility of the word in illustrative speeches.

We consider this method of syntagmatic “weighing” of words to be objective, since each lemma has theoretically equal chances to be represented in the dictionary by a phrase combination or illustrative speech. The more phraseological combinations with this lemma are presented in a dictionary article, the more its syntagmatic weight (S-weight) is. It is unlikely that the non-distinguishing of composite nominations and phrase combinations leads to errors in calculating the syntagmatic weight of a word: after all, composite nominations are also syntagmas, so the participation of a word in composite nominations should be taken into account when studying its syntagmatic activity. On the contrary, ignoring this circumstance can lead to a distorted view of the syntagmatic activity of a word expressed by its S-weight. See Table 2 for the distribution of Turkish vocabulary about C-weight.

The data of Table 2 is clearly presented on Figure 2.

As you can see from Table 2 and Figure 2, the compatibility in the source dictionary is worked out unevenly: there is a compact syntagmatic core of 1000–2000 words and an extensive periphery. 18.573 one-word lemmas out of 25.097 (which is 74% of the source dictionary) have no information about compatibility.

In the source dictionary, the syntagmatic dominant with 168 PhC is the noun el I ‘hand(s)’, and the syntagmatic vice-dominant is the noun iç ‘interior’ with 126 PhC. Next in descending order of the number of PhC are words, among which nouns predominate: su ‘water’ 113 ; yer ‘earth’ 105 ; iş ‘work, labor’ 97 ; ağız I ‘mouth, jaw’ 91 ; ayak ‘leg(s)’ 82 ; dil ‘tongue’ 82 ; yüz II ‘face’ 77 ; baş ‘head’ 74 ; üst ‘upper part’ 71 ; can ‘soul’ 67 ; Allah ‘Allah, God’ 63 ; kan ‘blood’ 63 ; söz ‘word, speech’; 59 ; kafa ‘head’ 58 ; akıl ‘mind’; 56 ; etmek ‘do’ 52 ; gönül ‘soul, heart’ 50 , etc. The syntagmatic core (S-core) of the source dictionary includes 1,213 words with at least three phraseological combinations. Sintagmatic nucleus , dominant and vice-dominant in the dictionarysource revealed for the first time.

Paradigmatic stratification of Turkish vocabulary

Paradigmatic stratification of vocabulary involves the identification of synonymous series from the smallest (2 words) to the largest (8 words). In order to implement the paradigmatic stratification of the Turkish vocabulary, a database containing a separate record of the interpretation (or Russian equivalent) of each individual meaning of each word was created. This is based on the assumption that a polysemous word can enter the synonymous series by any of its meanings, and the maximum number of synonymous series that includes the word is theoretically limited only by the number of its meanings.

Table 2. Distribution of Turkish vocabulary by S-weight

PhC	Words	Cumul.	S -weight
168	1	1	0,99996
126	1	2	0,99992
113	1	3	0,99988
105	1	4	0,99984
97	1	5	0,99980
91	1	6	0,99976
82	2	8	0,99968
77	1	9	0,99964
74	1	10	0,99960
71	1	11	0,99956
67	1	12	0,99952
63	2	14	0,99944
59	2	16	0,99936
58	1	17	0,99932
56	1	18	0,99928
54	1	19	0,99924
52	1	20	0,99920
50	1	21	0,99916
49	1	22	0,99912
47	1	23	0,99908
44	1	24	0,99904
42	2	26	0,99896
40	2	28	0,99888
39	1	29	0,99884
38	1	30	0,99880
35	2	32	0,99872
34	4	36	0,99857
33	1	37	0,99853
32	1	38	0,99849
31	2	40	0,99841
30	3	43	0,99829

1,20000

PhC	Words	Cumul.	S -weight
29	2	45	0,99821
28	1	46	0,99817
27	2	48	0,99809
26	5	53	0,99789
25	5	58	0,99769
24	2	60	0,99761
23	7	67	0,99733
22	5	72	0,99713
21	4	76	0,99697
20	9	85	0,99661
19	13	98	0,99610
18	4	102	0,99594
17	10	112	0,99554
16	10	122	0,99514
15	14	136	0,99458
14	20	156	0,99378
13	18	174	0,99307
12	17	191	0,99239
11	33	224	0,99107
10	41	265	0,98944
9	38	303	0,98793
8	55	358	0,98574
7	83	441	0,98243
6	109	550	0,97809
5	130	680	0,97291
4	198	878	0,96502
3	335	1213	0,95167

2	665	1878	0,92517
1	4646	6524	0,74005
0	18573	25097	0,00000

т5

5 g =  1,00000    -,•#•••••••••••••••••••••••••••••••••••••••••••••••••#•••6

о ° 0,80000  /

.У = 0,60000 I и ® 0,40000 I а Л             I

| У 0,20000 I

5 5 0,00000 •

^1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61

Number of phraseological combinations

Fig. 2. S-weight of words in the dictionary-source depending on the number of PhC

In total, there were 37,909 entries in the database. (Phrases were not included in the synonymous series: only one-word lemmas were taken into account.) The database rows were sorted by similarity of the right (Russian, i.e. metalanguage) parts. When definitions coincided, synonyms appeared in adjacent lines. Words which definitions coincide by 100% (at least in one of the meanings) were considered operationally potential synonyms. Meanwhile, not dictionary entries were compared, but definitions of lemmas presented in dictionary entries – lexico-semantic variants (hereinafter LSV). The number of meta words (Russian content words) in the definition is taken as 100%. Paradigmatic “weighing” of vocabulary, involving human participation, is the most time-consuming and least automated part of parametric analysis. However, when we adopt the strictest possible understanding of synonymy requiring 100% convergence of definitions, we risk losing some of the synonyms represented in the dictionary. For example, fetiş ‘amulet’; maskot ‘amulet’, and muska ‘amulet, talisman’; tılsım ‘amulet, talisman’. Formally, we get two two-member synonym series. If we reduce the threshold from 100% coincidence of definitions to 50%, then this will allow us to obtain a 4-member synonymic series combining all these words. Similarly, at 100% threshold we get 3 binomial synonymous series: 1) mecalsiz ‘powerless, infirm’; takatsiz ‘powerless, infirm’; 2) dingin ‘powerless, infirm; weak’; kudretsiz ‘powerless, infirm; weak’ and 3) güçsüz ‘powerless, weak, infirm’; kuvvetsiz ‘powerless, weak, infirm’. When the threshold is lowered to 50% and the restriction on the order of meta words is removed, all 6 words turn out to be synonyms. This approach may seem rough, but in most cases it gives a completely acceptable result, which can be considered as materials for a dictionary of synonyms. The massive and frontal nature of the dictionary survey inevitably leads to the approximation of semantic analysis. But the task of PAL is not to compile an impeccable computer dictionary of Turkish synonyms, but to get the paradigmatic weight of Turkish words, to “weigh” Turkish words according to the paradigmatic parameter.

Turkic languages, to which Turkish belongs, have their own specificity and, although the scientific validity of PAL has been repeatedly proven and tested, the application of PAL to each new type of language requires “adjustments to linguistic reality”, which we did when calculating F-weight, shortening the verbs with the affix -mak/ -mek. When researching the paradigmatic parameter of the vocabulary of any language (including Turkish) the following restriction is imposed on the concept of synonyms: words with different roots are recognized as synonyms. As a result of the solutions described above, “synonyms” with the same root are excluded from the series characterized by an operationally understood identity of semantics. Of all the word families in the formally identified synonymic series, one (as a rule, the shortest and least marked) word remains. A marked word is considered to have any restrictive or stylistic markings. The derived word, by the presence of an additional affix (and its inherent meaning), is marked in relation to the producing one. For example, from the synonymic series with the meaning ‘healthy’: esen, iyi, pürsıhhat, sağ, sağlam, sağlıklı, salim, sıhhatli, the lemmas sağlam, sağlıklı are excluded and one lemma sağ is left. The number of the synonyms changes, there is: not 8, but 6. Similarly to the synonymous series discussed above, the dimension of the synonymous series with the meaning ‘critical’ also changes: eleştirel, eleştirici, eleştirmeci, eleştirmeli, kritik, tenkidî, tenkitçi. The variants eleştirici, eleştirmeci, eleştirmeli are excluded, the shortest variant eleştirel is left. The tenkitçi variant is excluded from the tenkidî-tenkitçi pair. As a result, the number of synonyms meaning ‘critical’ is reduced from 7 to 3: eleştirel, kritik, tenkidî.

We take the dictionary source for granted by examining all the words presented in it. If there are markers in the dictionary that indicate the archaic and outdated nature of words, we can take them into account, if there are no such markers, we analyze what the dictionary gives. The application of the principles and approaches described above allowed us to obtain the results presented in Table 3.

The distribution of words by P-weight is clearly presented on Figure 3.

Paradigmatic dominant vocabulary is marked by a 8-member synonymous series in the source dictionary 4 (P-weight 0.99988): ‘strength, power’: çelim , erk , güç , kudret , kuvvet , mecal ,

Table 3. Stratification of Turkish vocabulary by the size of synonymous series

Words	Series	Cumul.	P-weight
8	1	1	0,99997
7	4	5	0,99986
6	5	10	0,99971
5	17	27	0,99922
4	79	106	0,99694
3	339	445	0,98714
2	1916	2361	0,93177
1	32241	34602	0,00000

1,20000

1,00000

0,80000

0,60000

0,40000

0,20000

0,00000

0,93177

0,98714

0,99694

0,99922

0,99997

Number of synonyms

0,99971 0,99986

0,00000

Fig. 3. Distribution of Turkish vocabulary by P-weight pehlivanlýk, zor. The paradigmatic vicedominants of Turkish vocabulary are represented by 7-member synonymous series with the meanings: ‘chest’ bağır, döş, göğüs, koyun, meme, sadır, sine; ‘sad’: hüzünlü, içli, kederli, mağmum, mahzun, pürmelâl, üzgün and ‘carefree’ : ferah, gailesiz, gamsız, geniş, kedersiz, meraksız, üzüntüsüz.

The words ferah and geniş in this synonymous series may seem foreign. If we take into account only their first meanings, this is indeed the case: they form their own synonymous series with the meaning ‘wide, spacious, roomy’, characterizing rooms, but not people. However, if we pay attention to their figurative meanings: “ ferah 2) figurative meanings ‘carefree, careless’; geniş 2) figurative meanings ‘carefree, careless’”, we will have to change our mind. This means that ‘latitude’ is transferred from physical space to the breadth of the human soul.

The paradigmatic vice-dominants of Turkish vocabulary are represented by 6-member synonymous series with the meanings: ‘healthy’; ‘in love’; ‘hashish’; ‘coquette’; ‘memory’. The

5-member synonymous series is represented by the meanings: ‘lightning’; ‘neutral’; ‘taste’; ‘pride’; ‘likeminded’; ‘earth’; ‘lowness, meanness’; ‘ordinary, mediocre’; ‘organ’; ‘offer’; ‘commitment’; ‘permission, approval’; ‘pimp’; ‘holiness’; ‘word’; ‘falcon’; ‘toilet, restroom’. Paradigmatic dominants in the source dictionary form the meanings of ‘strength, power’, ‘careless, carefree’, ‘chest’ and ‘sad, sorrowful’. Paradigmatic vice-dominants are 6-member synonymary series. The paradigmatic core (P-core) of the dictionary-source vocabulary consists of words that are included in all 2,360 selected synonym series.

Can you name the most important meaning in Turkish? The dictionary (Yusipova, 2005), treated with PAL, says: it is ‘strength, power’. This information is also received for the first time.

Epidigmatic stratification of Turkish dictionary

The epidigmatic depth of the source dictionary, measured by the maximum number of meanings, is 33 meanings, which is a lot for a dictionary of such size. The distribution of words by the number of meanings is presented in Table 4.

The most polysemous words which have from 13 to 33 meanings are verbs; polysemous nouns occur in the range of 2–12 meanings. Consequently, according to the dictionary-source, the superpolysemy is a characteristic of Turkish verbs, and the superphraseology (see syntagmatic stratification above) is a characteristic of Turkish nouns.

The data from Table 4 is clearly presented on Figure 4.

Figure 4 indicates that epidigmatics (polysemy) in the dictionary-source is worked out more evenly than (cf. Fig. 3) syntagmatics (compatibility).

The most polysemantic word in the dictionarysource (33 meanings) – E-dominanta – is the verb

Table 4. Distribution of words by the number of meanings in the dictionary (Yusipova, 2005)
Meaning	Words	Cumul.	E-weight	Example	Meaning
33	1	1	0,9999	çıkmak	go
29	1	2	0,9999	çekmek	pull, drag
26	1	3	0,9998	gelmek	to come
23	1	4	0,9998	çıkarmak	pull out, take out
21	1	5	0,9998	tutmak	hold on, hold
19	1	6	0,9997	almak	take
18	1	7	0,9997	yapmak	do; perform
16	2	9	0,9996	atmak; düşmek	throw, fall
15	1	10	0,9996	olmak	be, happen
14	2	12	0,9995	vurmak; kaldırmak	beat, hit; raise
13	3	15	0,9994	açmak; geçmek	open, move on
12	7	22	0,9991	kol	hand
11	6	28	0,9988	taban	sole, foot
10	14	42	0,9983	iç	inside, inside
9	25	67	0,9973	yüz	face
8	33	100	0,9960	baba	father, dad
7	52	152	0,9939	taş	stone
6	91	243	0,9903	boğaz	throat, throat
5	245	488	0,9805	ot	grass
4	536	1024	0,9592	öz	native (relatives)
3	1449	2473	0,9014	et	meat
2	5613	8086	0,6778	ay	moon
1	17011	25097	0,0000	aş	food

1,20000

1,00000

>0,80000

0,60000 I

0,40000

0,20000

0,00000

Number of meanings

Fig. 4. Dependence of E-weight words on the number of meanings in the dictionary source

çıkmak 1) ‘go’. The second most polysemantic word in the dictionary-source (29 meanings) – E-vice-dominanta – is the verb: çekmek 1) ‘pull, drag.’ We take 2473 words with 4 or more meanings as the epidigmatic core of the source dictionary. All this information about Turkish vocabulary was received for the first time.

Parametric stratification of Turkish lexicon

Now we have come to the culmination of our research, in which the particular parametric stratification of Turkish words in (Yusipova, 2005) develops into a monolith of the systemic stratification of Turkish vocabulary. Moreover, we get an idea of the system stratification of the Turkish dictionary and the place of each word in this stratification.

The integral parametric weight of words was calculated as follows. For each of the particular parameters, sets of words (about 1000 words) with the maximum particular weight were taken: F-core was 983 words (2–3 letters long), C-core was 1213 words (the number of PhC from 3 to 168), the E-core consisted of 1024 words (meanings from 4 to 33), the P-core was 2,360 synonym series (2–7 synonyms). The addition of particular weights for the words included in these sets gave a picture presented in Table 5.

Words with an I-weight, rounded to 4, make up the Small parametric core. Words with an E-weight, rounded to 3 or more, make up the Middle parametric core. Words with an I-weight, rounded to 2 or more, make up the Large parametric core of the dictionary. Finally, words with an E-weight, rounded to 1 or more, make up the parametric core of the dictionary-source. Words that do not get into the core according to any of the parameters, make up the system periphery of the dictionary.

The Small parametric core of the source dictionary contains the following 140 words (after the meaning the integral weight of the word is given): iş ‘work, labor’ 3,95731; üst ‘the upper part, the top’ 3,94826; ek ‘supplement, app, addition’ 3,94325; el II ‘stranger’ 3,93964; er I ‘man’ 3,93560; can ‘soul’ 3,93319; top ‘ball’ 3,93175; dil ‘language’ 3,91997; zor ‘difficulty’ 3,91953; iç ‘inside’ 3,91794; ruh ‘soul, spirit’ 3,91690; iyi ‘good’ 3,91115; almak ‘take’ 3,91053; düz I ‘smooth, even, flat’ 3,90814; ana ‘mother’ 3,89777; ak ‘white’ 3,89579; dip ‘bottom’ 3,89431; kaş ‘eyebrow’ 3,88531; ağız I ‘mouth’ 3,88178; baş ‘head(also figurative meaning)’ 3,88119; yüz II ‘face’ 3,87944; alt ‘bottom’ 3,87801; kol ‘arm’ 3,87769; bakmak ‘look’ 3,87753; ip ‘rope’ 3,87499; sıra ‘row’ 3,87492; tek I ‘the only one’ 3,87482; yapmak ‘do, make perform’ 3,87458; gelmek ‘to come, to arrive from somewhere’ 3,87343; dem I ‘breath, sigh’ 3,87196; tutmak ‘hold’ 3,87171; yan ‘side’ 3,87136; durmak ‘stand, be / remain motionless’ 3,86880; kalmak ‘stay’ 3,86402; kör ‘blind’ 3,86132; arka ‘back’ 3,86131; dış ‘external / exterior side, external / exterior appearance, appearance’ 3,86064; tam ‘full, whole’ 3,85897; kuru ‘dry’ 3,85895; baba ‘father, dad’ 3,85860; adam ‘person’ 3,85812; mal ‘property, state’ 3,85681; kök ‘root, rhizome’ 3,85522; çekmek ‘pull, drag’ 3,84741; vurmak ‘beat, hit’ 3,84701; yanmak ‘burn, light up’ 3,84482; dava ‘lawsuit’ 3,84243; gün ‘day’ 3,84016; hak I ‘rights’ 3,83594; asıl ‘base, basis’ 3,83334; boy II ‘height’ 3.83274; kırmak ‘smash, break’ 3,83247; kötü ‘bad’ 3,83242; ham ‘unripe (about fruit)’ 3,82805; usta ‘master, craftsman, expert in his field’ 3,82502; kalp I ‘heart’ 3,82425; ateş ‘fire’ 3,82302; dam I ‘roof’ 3,81976; ayrı ‘separate, detached’ 3,81513; dost ‘friend’ 3,81513; pis ‘dirty, stained’ 3,81458; sırt ‘spin’ 3,81361; ayak ‘leg, paw (animal) foot (insect)’

Table 5. Stratification of the vocabulary-source (Yusipova, 2005) by rounded integral weight

Sets				IntRound	R.R. Jusipova Dictionary
Dictionary	Large	Middle	Small	Weight	Words	Cumul.	AccNum
		Core	Core	4	140	140	0,56%
Core	Core		Periphery	3	490	630	2,51%
		Periphery		2	2604	3234	12,89%
	Periphery			1	3627	6861	27,34%
Periphery				0	18236	25097	100,0%

Note. IntRound – Integral, total parametric weight of words, rounded to whole; AccNum – accumulated number of words = lemmas in the database.

3,80493; orta ‘middle’ 3,80413; hava ‘air’ 3,80189; sıkı ‘tight, narrow’ 3,79863; âlem ‘world’ 3,79860; iğne ‘needle’ 3,79230; kese I ‘bag’ 3,79071; oyun ‘game’ 3,77923; askı ‘hanger, hook (for hanging clothes)’ 3,77313; kara II ‘black’ 3,76731; ağır ‘heavy’ 3,76660; yapı ‘building, construction’ 3,76301; sert ‘hard, solid’ 3,76270; ocak I ‘hearth, furnace, oven, stove’ 3,75823; yürümek ‘go, move, walk’ 3,75481; boya ‘paint’ 3,75086; kıyı ‘coast’ 3,74652; sulu ‘juicy’ 3,74134; küme ‘heap, pile’ 3,72010; düzen ‘order’ 3,67181; yatak ‘bed’ 3,66817; dünya ‘world, universe, earth’ 3,66814; çatal ‘fork’ 3,65547; fitil ‘wick, cord’ 3,65185; kabak ‘courgette, pumpkin’ 3,65113; resim ‘picture’ 3,65113; ciğer ‘lungs’ 3,64503; düşük ‘low’ 3,63775; hanım ‘khanim, khanum, mistress’ 3,63775; yavaş ‘slow’ 3,63775; kadın ‘woman’ 3,63208; sinir ‘nerve’ 3,63069; örnek ‘sample, model’ 3,62468; içeri ‘inside’ 3,61627; parti ‘party’ 3,61121; aşağý ‘bottom, bottom part’ 3,60523; çukur ‘pit, depression, excavation’ 3,60292; canlı ‘live’ 3,60009; zaman ‘time, period’ 3,60001; küçük ‘small’ 3,59901; karın ‘belly’ 3,59790; yağlı ‘fatty, oily’ 3,59698; karar ‘solution’ 3,59618; sıcak ‘heat’ 3,59618; doğru ‘Straight’ 3,59595; güzel ‘beautiful’ 3,59459; kanlı ‘bloodied, in blood’ 3,59459; çıkış ‘exit’ 3.59387; kızıl ‘bright red, red’ 3,59387; takım ‘group, company, circle of persons, team’ 3,59208; demir ‘iron’ 3,59184; kâğıt ‘paper’ 3,59184; şeker ‘sugar’ 3,59033; gedik ‘slit, crevice, crack’ 3,59025; tarak ‘comb’ 3,58869; duman ‘smoke’ 3,58814; kenar ‘Edge’ 3,58507; çevre ‘circumference’ 3,58049; kalın I ‘thick’ 3,58049; kanat ‘wing’ 3,58049; telli ‘fibrous’ 3,58049; kulak ‘ear’ 3,57817; hazır ‘ready’ 3,57411; kırık I ‘broken’ 3,56953; salma ‘let, let go’ 3,56953; pamuk ‘cotton’ 3,56897; kesme ‘slaughter’ 3,56742; oğlan ‘boy’ 3,56742; dalga ‘wave’ 3,56347; idare ‘management, guide’ 3,55407; tulum ‘waterskin’ 3,55407; bebek ‘infant, baby’ 3,54606; damla ‘drop’ 3,54606; rahat ‘rest, tranquility’ 3,54606; çalım ‘boasting, bragging, arrogance’ 3,53271; cephe ‘facade’ 3,53271; ortak ‘partner, companion, accomplice’ 3,53271; toprak ‘land’ 3,50124.

Since the purpose of PAL is to identify the cores of the lexical-semantic system of Turkish language, the consideration excludes lemmaphrases, which are means of secondary nomination, and only one-word lemmas are taken into account. Are we not distorting the real picture of the lexical-semantic system of language? No: the “Frequency Dictionary of Turkish Written Language” made on a sample of 1 million word-uses and numbering 22,693 words (Göz, 2003) contains 3,863 composite nominations – 17% of the total dictionary. At the same time, the total frequency of these nominations is 30,480 word-use. Consequently, composite nominations, covering only 3% of the Turkish text, are low-use peripheral units of Turkish vocabulary, and their exclusion from consideration cannot significantly affect the selection of the core of the Turkish language lexical and semantic system.

The author of the “Frequency Dictionary of Turkish Written Language” writes in the foreword: “This dictionary was created twice. The first study from early 1997 to the end of 1999 based on written publications was abandoned. Groups of 2 or 3 words (e.g. acil servis, a qk hava sinemas ) were counted as one unit there, while in the “Turkish Language Spelling Guide” of the Turkish Language Association (TLSG TLA) they were counted as the independent words acil , servis , a’k , hava , sinema . Therefore, we decided to start working again” (our translation. – Z. D. , A. K. ) (Göz, 2003). Recent borrowings from the English language and Greek-Latin internationalisms (for example, know-how , stand-by , post-scriptum ) were also excluded from further consideration. Turkish words written with a hyphen (for example, sıfat-fiil ‘ gram . the participle’ or tink-tank ‘ spoken bosses’) were taken into account during the analysis.

Since the purpose of the study is the cores of the lexical-semantic system of language, the consideration excludes words that do not carry the actual lexical semantics and are not primary names and verbs, including numerical, adverbs, pronouns, imitatives, predicatives and function words. In Turkish, there is a kind of parts of speech syncretism of adjectives and adverbs that differ not formally, but by their position in the sentence (compatibility). Words used not only as adverbs, but also as adjectives (e.g. hızlı ‘fast, impetuous, choppy’; ‘fast, impetuously, choppy’; ‘strongly, with all their might’) were included into the database, their adverbial meanings were taken into account.

We do not have statistics on the parts of speech in the Turkish dictionary, but in the dictionaries of the Russian standard language noun adverbs are 1.58% of the dictionary (Obratnyy slovar..., 1974, p. 944). In small Romanesque-Russian dictionaries, the representation of adverbs is as follows: in Romanian – 3%, in Italian – 2%, in Portuguese and French – 1% each and in Spanish – 0.48% [Titov, 2002, p. 186]. It is unlikely that in Turkish dictionaries these proportions are significantly different. So the exclusion of adverbs hardly damages the selection of the lexical-semantic core of the Turkish language.

Lexical semantics is concentrated in nouns, adjectives and verbs; adverbs borrow it from them through suffixation, reduplication, isolation, etc. Thus, the exclusion of adverbs from the lexical core in the parametric analysis of Turkish vocabulary cannot distort the lexical system of the Turkish language, also because the lexical semantics of the adverb is not independent, but is derived from names and verbs by which it will be presented. The adverb “ ok 1) a lot, 2) very, 3) long, 4) more than...” does not contain lexical semantics, performing LF (lexical function) Magn [Melchuk, Zholkovskiy, 1984]. Although this function is called “lexical”, it actually carries a grammatical meaning and refers not to vocabulary, but to the grammar of the language. In grammar it is impossible to do without it, in vocabulary – it is possible. Its antonym – the word az 1) insufficient, insignificant, meager, minuscule, 2) containing / having a small amount of something, 3) little, a little, 4) less” is taken into consideration, but not because it performs LF AntiMagn, but because it has the lexical meaning of the adjective: ‘insufficient, insignificant’. The range of parts of the speech is entirely determined by the interpretations taken by R.R. Yusipova in her dictionary. Predicatives var ‘there is, there are’, yok ‘there is not, none’, gerek ‘necessary’, lâzım ‘necessary’ have not lexical, but grammatical meanings: of the presence-absence or meaning of modality. It is illogical and impractical to include them in the lexical and semantic core of the language. Since the core of the lexical-semantic system is an appellative vocabulary, proper names ( onyms , as opposed to common names – appellatives ) are excluded from consideration , including ethnonyms – names of peoples, names of months, days of the week, letters, notes, etc.