Receptive intelligibility of Turkish to Iranian-Azerbaijani speakers

2.2.1. Ethnographic questionnaire

As it was mentioned, in order to control the role of non-linguistic factors such as language contact and negative attitude toward the Turkish language an ethnographic questionnaire was employed to select the appropriate participants. The questionnaire contained questions about the age and place of birth of the participants and the language that they speak at home, and some complementary questions to check the language contact and language attitude, such as if the participant had been to Turkey, if they listen to Turkish music, if they have taken any Turkish language course, if they watch or listen to Turkish TV and radio programs, and finally, what their attitudes toward the Turkish language and its learning are. Thus, through discarding the participants with prior language contact and negative attitude to the Turkish language (via the mentioned questionnaire), we controlled the non-linguistic factors which could have impacted the performance of the participants in the receptive intelligibility test. The questionnaire appears in Appendix 2.

2.2.2. Stimuli and task

As it was mentioned before, the main source of exposure to the Turkish language among Iranian Azerbaijanis is TV programs in Turkish, and consequently acquisition of Turkish language among Azerbaijanis seems to be achieved almost only by watching these Turkish TV programs. Therefore, we recorded some short videos from recently broadcast Turkish TV programs as the test stimuli. Using this kind of authentic material as data elicitation material in the process of the study, not only enhances the authenticity of the test in the present study, but also is considered a simulation of the real situation in which an Azerbaijani native speaker hears and reacts to the Turkish language. Therefore, the participants were asked to watch three short video clips recorded from Turkish TV programs twice; the first time they got a general understanding and the second time the video was played with pauses in the termination points of the sentences or phrases allowing them to translate what they had heard in Azeri. After translation they were asked to indicate the words which they found problematic, or words which impeded their comprehension. These words were identified as critical (unknown) words in this study.

In order to establish systematic and rigorous criteria for assessing the intelligibility level, the produced Azeri translations by the subjects were scaled based on the degree of correctness (Sağın-Şimşek & König, Citation2011); so that, the correct translations were coded “2,” partially correct translations were coded “1,” and incorrect translations or no translation at all were coded “0.”

In the second step, an attempt was made to get a deeper insight into either the sources of difficulty or the facilitating factors influencing the receptive intelligibility of Turkish to Azeri participants. The second part of the assessment was akin to “think aloud” process for identifying critical words. In this section, the subjects were asked to indicate the words which they found unknown and problematic. In this part, the proportion of known to unknown words to the majority of Azeri participants, or in other words, the percentage of critical words were used as the another factor influencing the degree of receptive intelligibility.

3.2.1. Lexical distance

Lexical distance in this study was calculated manually for each sentence, using the formula designed by Heeringa (Citation2004) and called “Levenshtein Distance.” According to the formula, in order to measure lexical distance the word pairs in aligned sentences were given points in accordance with their similarity: one point for non-cognates, half a point for a compound that is partly cognate, and zero points for cognates. Then, we calculated the correlation between lexical distances (in percent) for each sentence with the mean translation score (in percent) achieved for that sentence by the participants.

The total lexical distance (calculated through the mean of lexical distance for all sentences), was 12.45%. On the other side, in the whole text 19% of all the words were critical words which refer to the words that were not understandable to at least half of the participants. Words in this study were identified as critical for two linguistic reasons: first, because they were not part of the Azerbaijani lexicon or they did not have cognates in Azerbaijani and, second they were used in Azerbaijani in the form of an identical word or a cognate, but with a different meaning (i.e. false friends). In this regard, we had expected these critical words, at least partially, to block the flow of receptive communication. It is worth noting that we found the false-friends to have stronger negative effect on the flow of receptive intelligibility than unknown words; due to the fact that the false-friends were leading to a kind of misunderstanding which resulted from the participants’ false assurance of the word meanings. In addition, compared to unknown words, in the case of a false-friend, a person is not expected to use contextual cues to guess the meaning of the word, which in turn leads to the misinterpretation of the actual meaning and consequently, the phrase or sentence in which the false-friend is located. The correlation of (−0.63) which is significant at 0.01 level, shows significant effect of critical words, including non-cognates and false friends and those cognates that were not recognizable to the majority of the participants, in decreasing the level of intelligibility.

During the test, the participants were trying to find a similar word or cognate in their native language for the words they were hearing, based on the assumption that the both languages are similar with a difference in pronunciation. This effort can be beneficial or harmful depending on how it guides the person to find the correct corresponding or cognate in his/her lexicon, or misguides the person toward a similar word in his or her native lexicon with a different meaning. In this regard, the effect of false friends as a negative factor in receptive intelligibility was significant. As an instance, take into the account the sentences: bir köpük tabakası oluşturuyor, (it forms a layer of foam) and bir sümük tabakası oluşturuyor (it forms a layer of mucus). These sentences are identical except for the words köpük (foam), and sümük (mucus). Both words, köpük and sümük are also used in Azerbaijani, but while köpük means “foam” in both languages, sümük is a false-friend between these languages; sümük in Turkish means “mucus,” but it means “bone” in Azerbaijani. Thus, admittedly, the high difference in the mean translation score for the mentioned first and second sentences, can be attributed to the deleterious effect of sümük as a false-friend in the second sentence, which led all Azerbaijani participants to misinterpret or misunderstand the actual meaning. Similarly, in sentence (43) the critical Turkish word kend meaning “self,” was mistaken for the Azerbaijani phonologically identical word of kənd which means “village” by most of the participants.

In conclusion, it seems that Turkish and Azerbaijani share a significant number of false-friends, but the extent or degree to which these types of words can decrease the level of receptive intelligibility depends on many factors, among are which syntactic category or the part of speech and semantic role that the word plays in a sentence. This finding has implications for those Azerbaijanis and Turks who tend to learn Turkish and Azeri by the help of their own language.

As it was mentioned before, the grammatical category of a critical word in a sentence is another factor appearing to have influenced the degree of intelligibility. In a way, critical words in the verb, adjective, and noun position carrying a higher semantic load had the greatest negative impact on the receptive intelligibility, but those critical words with functional category or grammatical words had the minor or, sometimes even no role in deteriorating the level of understanding. According to what was observed in the present study, Turkish nouns and verbs, were the most critical syntactic categories either in facilitating or obstructing the flow of receptive intelligibility, while adjectives and adverbs should be put in the second place because the participant could guess the meaning of the these types of words. This factor can be attributed to the higher semantic load of the nouns and verbs compared to adjectives and adverbs which are used to modify the nouns and verbs. As an example in the sentence: “ekip ikinci dalış yerine gitmek için harekete geçtı,” (the crew started to go to the second diving place), the word “dalış” (diving) is a noun which is not used in Iranian-Azerbaijani and due to its significant semantic load proved problematic for most of the participants. While, the unknown adjective of “ufak” (small) in the utterence “ufak balıklar falan varmış orda” (there are small fishes over there), was not that problematic in terms of overall understanding. Furthermore, as mentioned by Öztopçu (Citation1993) “Turkish and Azeri due to a high typological approximity use mostly similar gramatical features such as agglutination, inflection of nouns for case, formation of passive, causative, reflexive, reciprocal, and negative verbal structures by suffixes, using postpositions instead of prepositions, having a subject–object–verb word order, modifiers preceding the modified head nouns, and absence of gender, definite article, duality, certain initial and final consonant clusters,” these features can be considered as one of the other facilitating factors in receptive multilingulism between these languages. The transcripts in Appendix 1. reveal how similar word order these languages use. Threfore, due to the fact that there is no need to learn a different gramatical structure, Azerbaijanis can focus mostly on learning the unknown words of Turkish to have a better understanding of that language.

Another point that is worth mentioning is that the same or similar words in these languages have a broader meaning in one language and a narrower meaning in the other. These types of words also have to be categorized as “false friends.” For instance, the widely used word of yapmak in Turkish, which means “to do, to make, to create, to practice, to build,” in Azerbaijani is exclusively being used in the meaning of “baking,” specially baking bread. The mentioned word which is frequently used in Turkish, as a false friend, confused most of the participants.

Although Turkish language is one of the languages which many Azerbaijanis in Iran are familiar with, the unofficial status of Azerbaijani language in Iran is one of the reasons for language attrition and slight deviation of it from its northern version which is spoken in Republic of Azerbaijan. Azerbaijani in Iran is influenced by Persian as the only official language in this country, so some Turkic words which are used in republic of Azerbaijan in northern Azerbaijani, in southern version in Iran have been replaced by Persian words. An example from the test is the word dalış (diving), which is used in the same form and meaning in the republic of Azerbaijan but almost not in southern Azerbaijani. Therefore, the expectation of a higher receptive intelligibility of Turkish to Azerbaijani speakers living in republic of Azerbaijan, where Azerbaijani is the official language of the country and the language of education, seems completely logical, due to relatively smaller lexical distance of that language from Turkish.

Turkish seems to have more English loanwords than Azeri, something which appeared to play a facilitating role or a shortcut for Azerbaijanis with knowledge of English in understanding the Turkish language. For example, some of the critical Turkish words which were loan words from English like komple (completely) and fotoraf (photograph), were identified by a large number of the participants who had at least a slight knowledge of English. So it seems that the participants in their first confrontation with Turkish language, applied all their linguistic knowledge derived from all other languages they knew. In addition, there were a significant number of cognate words in the test that were easily recognized by most of the participants, such as dişler, ağız, keyfı, ikinci, açık, soğuk, balık, korku, which some have either no or very low phonological distance or deviation from their Azeri counterparts. The identification of these keywords helped students to guess the meaning of the phrases and sentences.

It is clear that most of the critical words were problematic to the participants, because they were not a part of the Azerbaijani lexicon. However, it needs to be mentioned that synonyms of those critical words would not be necessarily unknown to Azerbaijanis. For example, the synonym of the critical Arabic-origin word of sene (year) in the study, is yil which is a cognate to Azerbaijani word of il. This shows even if some Turkish words are considered critical their synonyms are not necessarily unknown words for Azerbaijanis.

3.2.2. Phonetic distance

In another step an attempt was made to gage the role of phonetic distance as a factor which proved to influence the level of receptive intelligibility. In order to measure the phonetic distance of cognates of Azerbaijani and Turkish, we implemented the so-called “Levenshtein Distance” (Heeringa, Citation2004). The “Levenshtein” distance is measured based on the aligned words in equivalent sentences. The mentioned distance according to the formula was calculated between cognates in the test.

The “Levenstein” distance is measured based on the number of the symbols inserted, substituted or deleted, so the more the operations the more the phonetic distance will be. The following items are possible operations and their relevant distance:

•	Insertions and deletions 1 point,
•	identical symbols 0 points,
•	substitutions of a consonant by a consonant and a vowel by vowel 0.5 point,
•	substitutions of a vowel by a consonant or of a consonant by a vowel 1 point,
•	Diacritics were joined with the preceding symbol, adding an extra 0.25 point.

Diacritics were joined with the preceding symbol, adding an extra 0.25 point. So, for example, the distance between [a] and [a:] was 0.25, that between[a] and [o] 0.5, and that between [o] and [a:] 0.75. The unwanted effect of word length was compensated for by dividing the total sum of costs by the number of symbols aligned. In Heeringa (Citation2004) a more extensive explanation of the procedure is given.

Table 1 shows two cognates of Turkish and Azerbaijani and how phonetic distance is calculated according to phonetic variation.

Thus, phonetic distance for the mentioned cognates is the sum of the phonetic variation calculated for each alignment divided by the number of operations, which in the mentioned example the calculated distance is: (0.5 + 0.5)/6 = 0.16.

Azeri and Turkish use similar alphabetic sound letters or alphabetic signs which could add to the reading and writing intelligibility of these languages. Turkish and Azerbaijani alphabets are nearly identical. Turkish alphabet has 29 and Azerbaijani 32 characters. Thus, Azerbaijani has three more characters of x, ə, and q which do not exist in Turkish. So, while, Azerbaijanis are familiar with all the signs of Turkish, Turks need to learn these three more characters in order to be able to read Azerbaijani texts. These additional three characters correspond to particular Turkish sounds or signs, x for h, ə for e, and q for k in cognates or historically related words and also the common loanwords (Özel, Citation2008).

In the present study, the effect of phonetic distance on the receptive intelligibility was investigated through running a correlation between the calculated phonetic distance of each Turkish and Azeri sentence and the achieved translation scores by the participants for the relevant sentence. The result of correlation was (−0.215) which is not statistically significant. Thus, regarding the low-phonetic variation between cognates in the present study which in average was (4.7%), and the low correlation between phonetic distance and translation scores, we came to this conclusion, that the role of phonetic distance between Turkish and Azeri is not as significant as that of Lexical distance in determining the degree of receptive intelligibility. In this case, it needs to be mentioned that the result of this study is not in consistency with those carried out by Gooskens (Citation2006, Citation2007), in which phonetic distance rather than lexical distance was found to be the stronger predictor of receptive intelligibility between Scandinavian languages. Table 2 gives a deeper insight into the sound and phonetic variation between cognates of Turkish and Azeri.

Table 2 shows some examples of sound changes between cognates of Turkish and Azeribaijani. Turkish and Azerbaijani share a good number of cognates which differ in terms of pronunciation in one or more sounds and this variation, to a large extent is systematic and predictable. Some cognates differ in just one letter, for example, Turkish “bak”/ba:k/ for Azeri “bax”/ba:x/ in which velar/k/ changes to fricative/x/. However, in this study, we faced some words which despite their relatively great similarity to their Azerbaijani counterparts, were not successfully identified by most participants. Some examples from the transcripts are Turkish “defa” for “dəfə” and “aslinda” for “əslində” which implies that identification of that group of words within which Turkish vowel a/a/ changes to Azeri vowel ə/æ/ is one of the most difficult categories for identification. Thus, the similarities are not all the time a guarantee for word identification.

The point which is worth mentioning is that Turkish and Azeri share a significant number of Arabic and Persian origin loanwords, but mostly with a different pronunciation. These loanwords in Azerbaijani are more similar to their original forms than in Turkish. In other words, common loan words in Turkish have gone through a more drastic deformation, maybe, due to the fact that phonological system of Turkish has fewer phonetic features compared to Azerbaijani, and consequently less compatible with Farsi and Arabic phonetic systems.

Furthermore, the stronger deformation that common loanwords of these languages in Turkish have gone through was not expected by Azerbaijani participants. This kind of variation in pronunciation is based on a rigorous criterion which makes it easier to adapt to by Azerbaijanis after a time of exposure, when they are learning or trying to understand the Turkish language.