Chapter I. Lexical typology is the systematic study of cross-linguistic variations

2.2. Semantic categories of lexical typology
It is known that the syntactic analysis of long and complicated sentences is more difficult than the analysis of short sentences. A parsing success depends among other things also on the length of the input sentence. The idea of cascaded parsing described in some works is also used for the simplifying the translation process. The advantage of working with a cascade of specialized parsers instead of one very complex general parser is quite obvious – the complexity of the task is substantially reduced and the parsing process is speeded up. The syntactic analyzer of Azerbaijani developed within the frame of this MT system is considered for syntactic analysis of simple sentences. In order to translate the composite sentence (the complex and compound sentences are meant), firstly the composite sentence should be divided into simple sentences that compose this sentence. Hereinafter, dividing the complex sentence into the principal and subordinate clauses will also be referred as dividing the complex sentence into simple sentences. Then these simple sentences should be syntactically analyzed and after translating each simple sentence separately the translation of the composite sentence must be synthesized. Such approach helps to perform more qualitative syntactic analysis and consequently the translation quality considerably improves.
While dividing the composite sentence into the simple sentences it is necessary to answer three questions: 1) How to divide the composite sentence formally into simple sentences, 2) Is it possible to get the translation of a given composite sentence by dividing this sentence into the simple ones, translating the received simple sentences separately and “merging” these translations into one composite sentence? 3) Can there be any misinterpretation in the sentence after this process? To answer these questions let’s consider the types of composite sentences in and their origin features. Hereinafter, will be referred as the source and English as the target language. In (like in other languages) the composite sentences are divided into two groups according to peculiarities of their formation: 1) Compound sentences; 2) Complex sentences. Compound sentences are formed by connecting two or more simple sentences with comma or with coordinating conjuctions (“və” – “and”, “amma, ancaq, lakin” – “but”, etc.). In this case each simple sentence composing the compound sentence has a predicate. For example. “Buludlar səmanı bürüdü, hava soyudu və qar yağmağa başladı” – “The clouds covered the sky, it became cold and it began to snow”. In this sentence the simple sentence – “Buludlar səmanı bürüdü” – “The clouds covered the sky” is connected to the simple sentence – “hava soyudu” – “it became cold” with a comma and to the simple sentence – “qar yağmağa başladı” - “it began to snow” with conjunction “və” – “and”. Each of these sentences has their own independent predicate “bürüdü”, “soyudu” and “başladı”. In complex sentences one of the simple sentences is the main (principal) and the other one is of second importance (subordinate). The subordinate clause explains the principal clause from different points of view.
The complex sentences in turn are conditionally divided into two types in: 1) Analytic type; 2) Synthetic type. In analytic type of complex sentences the subordination of the subordinate clause to the principal clause is not so strong. That’s why the subordinate and principal clauses composing these kinds of complex sentences can be used as the independent sentences, in other words, each of them has a predicate. These sentences are connected to each other with subordinating conjunctions “ki” – “that”, “çünki” – “because”, etc. For example: “Mən evdən çıxmaq istəyirdim ki, dostum bizə gəldi” – “I just wanted to leave home when my friend came to us”. In this sentence both of simple sentences – “Mən evdən çıxmaq istəyirdim” - “I just wanted to leave home” and “Dostum bizə gəldi” – “my friend came to us” – has its own predicate (the wordforms “istəyirdim” and “gəldi”) and this criterion can be used in dividing the sentence into simple sentences. In synthetic type of complex sentences the subordinate clause can not be used separately as a simple sentence because the subordinate clause doesn’t have a predicate. As an example we consider the case when the subordinate clause is connected to the principal clause by particle-suffixes “-sa” or “-sə”. Other variants can be considered analogically. For example: “Kim çempion olmaq istəyirsə, o, yorulmadan məşq etməlidir” – “If somebody wants to be a champion, he has to train tirelessly”. In this complex sentence even though the subordinate clause “Kim çempion olmaq istəyirsə” doesn’t have a predicate the obvious criterion that connects this sentence to the principal sentence is the word-form istəyirsə. The particle-suffix “-sə” of this word-form “istəyir-sə” connects the subordinate clause to the principal clause (if we ignore this suffix, the word-form “istəyir” becomes the predicate of the subordinate clause) and it allows dividing this complex sentence into simple sentences: “Kim çempion olmaq istəyir” – “Somebody wants to be a champion”; “O, yorulmadan məşq etməlidir” – “He has to train tirelessly”. In this example we will also call a predicate the word-forms that adopt the particle-suffix “-sa” or “-sə” connecting the subordinate clause to the principal clause because such word-forms are not encountered in any other situation.
Thus, we come to a conclusion that it is possible to divide both complex and compound sentences into simple sentences in Uzbek. As mentioned above there is a word-form considered as a predicate in each simple sentence composing the composite sentence according to which the simple sentences can be defined? Furthermore, as mentioned above there can also be the separators such as comma and / “or” conjunction between the simple sentences. But one additional reason - the word order should be taken into consideration for application of these criteria, like in all Turkic languages, the word order in the sentence is free from the standpoint of location of the sentence members. For example, if we change the word order in the following sentence – “Mən universitetə’ gedirəm” – “I go tothe University) – all new sentences received (the number of such sentences is 6) will express the same meaning. In the process of formal syntactic analysis of the composite sentences the independence of word order can result in serious difficulties. For instance: “Böyük gəmilər okeanda üzə bilir, katerlər, qayıqlar isə sahildən çox uzağa gedə bilmirlər” – “Big ships can sail in the ocean, but launches, boats cannot go far-away from the shore”. This sentence can be divided into two simple sentences according to the predicate “bilir” and the comma given after this predicate. But if the word order in this sentence is like – “Okeanda üzə bilir böyük gəmilər, katerlər, qayıqlar isə sahildən çox uzağa gedə bilmirlər”, it’s very difficult to divide it formally into simple sentences as the predicate of the first simple sentence “bilir” doesn’t stand in the end of the sentence. However, if we don’t take into consideration the literary texts (particularly the poems), in real texts the predicate of the sentence usually comes in the end and the subject+object+predicate (SOV) topology is observed. Hereinafter, we are going to consider only the composite sentences relevant to this topology. Thus, the results obtained allow us to assume that the following conditions are met while dividing the composite sentence into simple sentences in Uzbek.
The increasing tendency to use semantic agreement with constituent types to the right, and especially with anaphoric pronouns, suggests that, with increasing syntactic distance from the controller, form fades in the memory of the speaker with meaning trumping it. Thus, in the case of German das Mädchen, the neuter form of the noun calls for neuter agreement on articles, demonstratives, adjectives, and relative pronouns that are placed close to the controller noun but the anaphoric pronoun, which is at some distance from it, picks up on the natural feminine gender of the referent. Regarding (a): while most languages that have indirect-object–verb agreement do indeed also have direct-object–verb agreement, this is to be interpreted as a paradigmatic generalization, not a syntagmatic one. That is to say, in ditransitive sentences (i.e. those with both direct and indirect object), verb agreement with the indirect object may take precedence over the direct object. This is true for Lebanese Arabic as shown in (29): while in this language, there is both direct-object and indirect-object agreement, there are ditransitive sentences where the verb agrees with the indirect object but not with the direct one. What might be the explanation for the prevalence of subject agreement over direct-object and indirect-object agreement?
The historical origins of verb agreement shed light on this issue. Verbagreement markers evolve from personal pronouns that refer to topicalized noun phrases. An example of this construction is English John, he is a friend of mine. In some languages (e.g. Arabic or Swahili), agreement markers still carry their anaphoric pronominal function: the noun phrases they refer to are optional. This anaphoric function may get lost over time agreement markers come to redundantly re-iterate some characteristics of a noun phrase that is also present in the sentence. If the origin of verb agreement markers is tied to topicalization, we can expect verb agreement to occur primarily with noun phrases that are most likely to be topicalized. Foremost of these are subjects: topicalization favors noun phrases that are definite and subjects of transitive sentences generally are. This suggests that verbs should most commonly agree with subjects – a true fact. Direct objects are not generally definite
but there is a telling pattern: if in a language the verb agrees with indefinite objects, it also agrees with definite ones. This is so in Lebanese Arabic: the examples in (28) show verb agreement with definite direct objects (‘the boy,’ ‘to the girl’) but if the object is indefinite, there is no verb agreement with it. The fact that in a given sentence, verb agreement with indirect objects takes precedence over agreement with direct objects (as in (29) above) may have to do with indirect objects being animate while direct objects mostly are not. Animate noun phrases are often more topical than inanimate ones. To summarize so far: Ross linguistically preferred verb-agreement controllers line up along the Controller Hierarchy given in GEN-5 (which, as you may have noticed, is a subpart of the Accessibility
Furthermore, there is also a second cross linguistically valid ranking of agreement patterns which constrains not agreement controllers but the agreement features that various agreement targets exhibit. Formulated by Greville Corbett, the Agreement Hierarchy can be illustrated as follows.