The Similarity and Mutual Intelligibility between Amharic and Tigrigna Varieties Tekabe Legesse Feleke Verona Univerisity Verona, Italy [email protected] Abstract without major difficulties (Demeke, 2001; Gutt, 1980). However, the similarity among the lan- The present study has examined the sim- guages is often obscured by the attitude of the ilarity and the mutual intelligibility be- speakers since language is considered as a sym- tween Amharic and two Tigrigna vari- bol of identity (Lanza and Woldemariam, 2008; ties using three tools; namely Levenshtein Smith, 2008). Hence, there are cases where vari- distance, intelligibility test and question- eties of the same languages are considered as dif- naires. The study has shown that both ferent languages (Hetzron, 1972; Hetzron, 1977; Tigrigna varieties have almost equal pho- Hudson, 2013; Smith, 2008). Therefore, due to netic and lexical distances from Amharic. politics, sensitivity to ethnicity and the lack of The study also indicated that Amharic commitment from the scholars, the exact number speakers understand less than 50% of the of languages in Ethiopia is not known (Bender and two varieties. Furthermore, the study Cooper, 1976; Demeke, 2001; Leslau, 1969).Fur- showed that Amharic speakers are more thermore, except some studies for example, Gutt positive about the Ethiopian Tigrigna va- (1980) and Ahland (2003) cited in Hudson (2013) riety than the Eritrean variety. However, on the Gurage varieties, and Bender and Cooper their attitude towards the two varieties (1971) on mutual intelligibility of Sidamo dialects, does not have an impact on their intelli- the degree of mutual intelligibility among various gibility. The Amharic speakers’ familiar- varieties and the attitude of the speakers towards ity to the Tigrigna varieties seems largely each others’ varieties has not been thoroughly in- dependent on the genealogical relation be- vestigated. Hence, the present study examined tween Amharic and the two Tigrigna vari- the distance and the mutual intelligibility between eties. Amharic and two Tigrigna varieties together with Keywords: Language Similarity, Lan- the effect of the attitude on the mutual intelligibil- guage Distance, Mutual Intelligibility, At- ity. titude, Language Contact Amharic and Tigrigna are members of the Ethiosemitic language family, a branch of proto- 1 Introduction Semitic family (Bender and Cooper, 1976; De- meke, 2001; Hetzron, 1972; Hetzron, 1977). 1.1 Language in Ethiopia According to Demeke (2001), Hetzron (1972), More than 85 languages are spoken in Ethiopia Hetzron (1977) and Bender and Cooper (1971), (Demeke, 2001; Hetzron, 1972; Hetzron, 1977; Ethiosemitic languages are divided into North and Hudson, 2013). The languages are classified un- South Ethiosemitc. While the Tigrigna varieties der four language families: Semitic, Cushitic, are North Ethiosemitic languages, Amharic is one Omotic and Nilo-Saharan (Bender and Cooper, of the South Ethiosemitic languages. Nowadays, 1976; Demeke, 2001; Hornberger, 2002; Hudson, Amharic is spoken only in Ethiopia, but Tigrigna 2013). In each family, there are many related lan- is spoken both in Ethiopia and in Eritrea. Due guage varieties so that the speakers of one vari- to the genealogical and typological relationship ety can sometimes communicate with the speak- between Amharic and Tigrigna (Demeke, 2001; ers of another variety in the same language family Hetzron, 1972; Hetzron, 1977), Amharic speak- 47 Proceedings of the Fourth Workshop on NLP for Similar Languages, Varieties and Dialects, pages 47–54, Valencia, Spain, April 3, 2017. c 2017 Association for Computational Linguistics ers are supposed to understand the Tigrigna vari- have been reported by the studies conducted on eties to a certain degree. Since Amharic has been the Scandinavian languages and the Chinese di- the national language of Ethiopia, it is a widely alects in this regard (see (Gooskens and Heeringa, used language compared to Tigrigna (Getachew 2004; Gooskens, 2013; Gooskens, 2007; Tang and and Derib, 2008; Iyob, 2000; Lanza and Wolde- Heuven, 2007; Tang and Heuven, 2009; Tang and mariam, 2008; Smith, 2008). The use of Amharic Heuven, 2015)). The present study is an addition as a national language helped many speakers of to these contributions. Ethiopian Tigrigna to learn Amharic as a second language (Smith, 2008). Moreover, Amharic has 1.2 Measuring Language Distance and also been given as a subject for Ethiopian Tigrigna Mutual Intelligibility speakers, starting from elementary school. Some The study of the distance among related languages speakers of Eritrean Tigrigna variety used to speak has been a concern of many scholars for decades Amharic before secession. However, after the in- (Sokal, 1988). Several previous studies employed dependence, using Amharic in schools and in dif- phonetic distance to measure the relative distance ferent offices was banned (Hailemariam and Wal- between various languages (Bakker, 2009; Cohn ters, 1999; Rena, 2005). The relationship between and Fienberg, 2003; Kessler, 1995). However, the peoples of the two countries was also strained the emergence of the Levenshtein algorithm has especially after Ethio-Eritrean war from 1988 to enhanced the objective structural comparisons by 2000. Hence, due to the border conflict, Eritrean introducing a computer-based distance computa- Tigrigna speakers do not also have an access to tion (Heeringa, 2004; Gooskens and Heeringa, Tigrigna speakers in Ethiopia and to the Amharic 2004). This has probably contributed a lot in terms speakers. of attracting many scholars towards the study of language variation (Gooskens, 2013). Recently, Studies on the language attitude of the speakers several studies have been conducted on European of Amharic and the Tigrigna varieties are at scarce. languages and on Chinese dialects, for example, However, language, ethnicity and politics are very (Gooskens and Heeringa, 2004; Heeringa, 2004; interrelated in Ethiopia (Bulcha, 1997; May, 2011; Tang and Heuven, 2007; Tang and Heuven, 2009; Smith, 2008). The link has been accelerated by Tang and Heuven, 2015) by employing the Lev- the ethnic-based federal system in Ethiopia (Lanza enshtein algorithm together with the mutual intel- and Woldemariam, 2008; Young, 1996; Vaughan ligibility and perceptive distance measures. For and Tronvoll, 2003).The atmosphere of politics in instance, Gooskens (2007) compared data from Eritrea and Ethiopia could also affect the attitude Scandinavian languages (Danish, Swedish and of the people in both countries. There has been an Norwegian) with that of West Germanic languages anti-Ethiopia sentiment in Eritrea since 1993 (Ab- (Dutch, Frisian and Afrikaans) and reported that bink, 2003; Assefa, 1996; Iyob, 2000). This hos- mutual intelligibility can be predicted based on tile situation could have an effect on the attitude phonetic and lexical distances. Similarly, Bezooi- of the speakers of Amharic and the speakers of the jen and Gooskens (2007) investigated the intelli- Ethiopian Tigrigna. gibility of written Afrikaans and Frisian texts for The study of the similarity between Amharic Dutch speakers and reported the association be- and the Tigrigna varieties and the attitude of the tween the Levenshtien distance and mutual intelli- speakers of one language towards another has a gibility. Heeringa (2004) also employed the Lev- paramount significance in two ways. From prac- enshtein distance for the comparison of Dutch and tical point of view, there has been an attempt to Norwegian varieties. standardize Tigrigna and use it widely in media The subjective measures often include percep- and in schools. The study positively contributes tive distance and functional tests (Gooskens and to this effort. From theoretical perspective, there Heeringa, 2004; Tang and Heuven, 2007; Tang have been a number of attempts towards improv- and Heuven, 2009). According to Gooskens ing the enduring limitations of methods of di- (2013), functional intelligibility between related alectology. One of the positive contributions has languages can be measured by employing con- been complementing the traditional lexicostatis- tent questions, translation, recorded text testing, tics methods by the mutual intelligibility and per- observations and performance tasks. Tang and ceptive distance measures. Very promising results Heuven (2009) employed word intelligibility test 48 and word recognition in a sentence to examine and Gooskens (2007) also used questionnaires to the mutual intelligibility among the Chinese di- examine the language contact and language back- alects. According to Gooskens (2013) and Tang ground of their participants. and Heuven (2009) , opinion test can be designed without speech. For example, speakers of a cer- 2 Research objectives tain variety can be requested to give their judg- The study was conducted to address, among oth- ment on the speakers of other varieties who live ers, the following four specific objectives. 1) To in certain geographical areas (estimated linguistic determine the distance between written Tigrigna distance). Bezooijen and Gooskens (2007) used varieties and Amharic. 2) To determine the atti- cloze test to measure the functional ineligibility tude of the native Amharic speakers towards the of written Afrikaans and Frisian for the native Tigrigna varieties. 3) To identify which Tigrigna speakers of Dutch. Swarte and Gooskens (2014) variety is more intelligible for the native speak- employed a word translation to measure the im- ers of Amharic.