Multilingualism and Local Literacy Practices in Ethiopia: Language contact in regulated and unregulated spaces1

The study of the linguistic landscape has provided a new dimension to theories and issues related to multilingualism, including language policy. In this growing field of inquiry, however, not enough attention has been given to the linguistic landscape in sites in the Global South. Since one of the aims of literacy studies is to reveal the variety and social patternings of practices, there is a need to compare linguistic landscape data with other various textual materials. In this article, we present linguistic landscape data from two federal regional capitals in Ethiopia that demonstrate multilingual language use. We also compare the linguistic contact patterns with those found in schoolbooks used in the same region. Such a comparison involves language use in unregulated as well as in regulated spaces (see Sebba 2009). Regional ethnically based languages are now being used in new arenas, including the linguistic landscape and education because of a new language policy promoting the use and development of regional languages. The two regional capitals provide privileged sites for examining the products of local literacy practices, involving values, attitudes, ideologies, and social relationships. We discuss the results in light of various ideologies and argue for the speaker-writer’s active mobilisation of multilingual resources in new language arenas.


IntroductIon
T he study of written language has come to the fore in current approaches to multilingualism. As a result, a 'new approach to multilingualism', the study of the linguistic landscape (Gorter 2006) or the written language in the public sphere has emerged. First proposed by Landry and Bourhis (1997) as a barometer for measuring ethnolinguistic vitality in Canada, this study is today a thriving field of inquiry documenting various socio-cultural aspects of languages in multilingual societies (Shohamy and Gorter 2008). In more recent work on linguistic landscape, the focus has been on signage in world cities and its role in the construction of social and cultural meaning in multilingual urban spaces (Ben-Rafael, Shohamy and Barni 2010). However, not much attention has been given as yet to the study of the linguistic landscape in urban sites in the Global South (but see Kasanga 2010;Reh 2004, Stroud andMpendukana 2009). Dagenais, Moore, Sabatier, Lamarre and Armand (2008) make particular mention of the paucity of studies on the linguistic landscape in the educational domain. Showing its value, Cenoz and Gorter (2008) discuss how the linguistic landscape can be seen as an additional source of input in second language acquisition, contributing to a type of literacy that is multimodal and multilingual.
In this article, we attempt to place the study of the linguistic landscape within a larger framework of literacy practices. We do so by specifically examining multilingual language use, particularly language contact as documented in the linguistic landscape of two regional capitals in Ethiopia, and by further comparing it with language contact phenomena in other texts, namely children's schoolbooks. Lüpke (2010:2) points out that 'language contact between African languages is dramatically underresearched', as most studies of language contact situations involving African languages entail contact between a local and a former colonial language. While work on language contact often deals exclusively with structural properties of language, a more functionalist perspective, as noted in Matras (2009: 4), rests on a view of language as social activity for which 'bilingual (or multilingual) speakers have a complex repertoire of linguistic structures at their disposal'.
Literacy is conceived of as 'situated social practices embedded within relations of culture and power in specific contexts' (Prinsloo and Baynham 2008:2). As Barton and Hamilton (2000: 7) point out: 'The notion of literacy practices offers a powerful way of conceptualising the link between the activities of reading and writing and the social structures in which they are embedded, and which they help shape.' Literacy practices involve values, attitudes, ideologies, and social relationships -in sum, how people in a particular culture construct literacy, how they talk about literacy and make sense of it. The social approach to literacy as articulated in Barton and Hamilton (2000), Barton (2001), and Prinsloo and Baynham (2008) emphasises the historical dimension to literacy. Literacy practices change and new ones are often acquired through processes of informal learning as well as through formal education and learning. The two regional capitals of Ethiopia in question provide privileged sites for examining the products of multilingualism and local literacy practices. Because of a new language policy promoting the use and development of ethnically based regional languages, these languages are now being used in new public arenas, including the linguistic landscape and education. Ethiopia does not have a colonial past yet English is an official second language and thus has a prominent place in the linguistic profile of the country. Hence a focus on regions in Ethiopia provides an opportunity to investigate language contact between African languages as well as contact with English, a language usually associated with globalisation, which has been recontextualised as a local language in Africa (Higgins 2009). In this article, we explore the issue of language contact in local literacy practices in Ethiopia, a country considered marginal from a global perspective. Focusing on regional capitals, we also take further steps to the margins within the country. Yet the urban areas in focus in each region are indeed centres within the margin. Such a vantage point allows us to trace multilingual literacy practices across trajectories of centre and periphery, or margin, within one country and within a global perspective.
In the following section, we first present an overview of the linguistic situation in Ethiopia and the new language policy that became operative in the early 1990s; this overview is contextualised within an historical framework. Next, we present the methodology we employed in collecting the various data in the two regions. Subsequently, we provide an analysis of the language use and particularly the language contact we found in the two types of data -the linguistic landscape and the educational texts -in light of local ideologies. In conclusion, we offer alternative interpretations of the patterns we found and discuss their implications for multiling-ualism on the margin.

MuLtILInguAL EtHIoPIA
E thiopia, located in the Horn of Africa, has a population of about 80 million and is multilingual, multiethnic and culturally pluralistic. According to Ethnologue (Lewis 2009), there are eighty-five living languages spoken, divided among four different language families: the Semitic, Cushitic and Omotic families of the Afroasiatic Phylum, and those belonging to the Nilo-Saharan Phylum (for an overview of languages in Ethiopia, see Crass and Meyer 2008). The Semitic languages, such as Amharic and Tigrinya, are spoken in northern, central and eastern Ethiopia. The Cushitic languages are mostly spoken in central, southern and eastern Ethiopia. The language of the largest ethnic group in the country, Oromo, is considered by Ethnologue (Lewis 2009) as a 'macrolanguage', that is, 'multiple, closely related individual languages that are deemed in some usage contexts to be a single language'.
Ethiopia's major ethnic groups include the Oromo, who speak the Cushitic language of the Cushitic people who make up about 40% of Ethiopia's total population. The Semitic Amhara and Tigrayans comprise only 32% of the population; however, historically they have dominated the country politically. Despite the common Semitic background of the Amhara and the Tigrayans, their languages are mutually unintelligible. Amharic diverges significantly from the other Semitic languages of Ethiopia as a result of its extensive contact with Cushitic and Omotic languages (Yimam 2004).
Amharic is used as a lingua franca by all Ethiopians. This practice is encouraged by a range of opportunities, including its function as a working language of the federal government, trade, urbanisation, labour migration, displacement and other forms of migration, education and literacy, and perhaps most significantly, by the high proportion of intermarriages between members of different ethnic or linguistic groups. Amharic continues to spread, both as a first language and as a second language, in spite of the introduction of regional languages for wider purposes (Cohen 2006: 171). Amharic continues to be the most widely used language in Ethiopia and has a long literary tradition. The sociopolitically dominant social group in Ethiopia has historically used and promoted Amharic while in fact, not all Amhara or all Amharic speakers were dominant or powerful. Nevertheless, the historical hegemony of the Amharic language, which has led to a form of domination, often regarded as ethnic in character, has been well documented. Some historians claim that the Amhara led the so-called 'Abyssinian Conquest' in the nineteenth century during which they 'colonised' territories to the south and west of the country, particularly the area predominantly inhabited by the Oromo (Marcus 2002). This claim has been disputed by other historians who prefer to describe the process as 'nation-building'. Nevertheless a strict policy of 'amharization' of the new regions ensued, actively promoted by the state under Emperor Haile Selassie's regime . Mazrui (2004: 6) points out that sub-Saharan Africans are rarely 'strong linguistic nationalists (such as in Asia, the Middle East and Europe, for example) (who) tend to resist any massive dependence on languages other than their own'. However, he notes that among the exceptions to this phenomenon are the Amhara, along with the Somali and the Afrikaners.
The policy of amharisation was based on the assumption that the use of one language would be necessary to produce national unity. Haile Selassie's government came to a dramatic end with the overthrow of the regime by a military coup and the transfer of power to a Soviet-backed communist junta, referred to as the Derg. In contrast with the policy during the former regime, there were some attempts at enhancing the status of regional languages during the Derg regime. In order to conform to the ideals of socialism and to demonstrate political change, the military government of the Derg attempted to use 15 regional languages as part of the national literacy campaign. However, while the government had declared that regional language development would be its aim, regional languages continued to be restricted to orality. Hence, the use of Amharic as the most prestigious language, particularly in literacy, continued nationwide.
In 1991, the country underwent a dramatic change in regimes and several major political, social and economic changes came about at the same time (Pausewang, Tronvoll and Aalen 2002;Smith 2008). The Ethiopian People's Revolutionary Democratic Front (EPRDF), the ruling political coalition in Ethiopia with the Tigrayan People's Liberation Front (TPLF) as the dominating party, was responsible not only for the overthrow of the Derg, but also for some important political changes. A new constitution advocating a policy of ethnic federalism was initiated. In terms of this, Ethiopia's Federal Constitution guarantees that persons belonging to various ethnic and linguistic minorities shall not be denied the right to enjoy their own culture and to use their own language. Various proclamations have been made to bring about the decentralisation of decision-making from central to regional administrations. Today there are nine autonomous federal regions, ethnically based, and two chartered cities, including the capital Addis Ababa.
Until 1991 Amharic was used as the language of instruction and literacy in primary education. After the downfall of the communist military regime, the newly formed government introduced a national educational policy based on the use of 'mother tongues' as the medium of instruction in all public schools. The stated purpose of the policy was to foster national unity, identity, and development while respecting cultural diversity. Yet the implementation of the educational policy shows that the regionally dominant language serves as the official working language and language of education throughout the region, which is also the homeland of other minority linguistic groups. Hence 'mother tongue' meant in reality and still generally means the regional language. This language policy of ethnic federalism as applied to education has its critics who argue that education in the regional languages will inhibit social and national mobility (Vaughan 2007).
The policy of ethnic federalism was mainly introduced to recognise the various ethnolinguistic groups in the country, and the goal of using ethnic/regional languages was to satisfy the diverse needs of Ethiopia's multiethnic and multilingual population. There was, however, no clear statement in the language policy about how this goal should be attained. In practice, this was achieved by the decisions taken by official bodies. Today regional and local languages are widely used in the educational, administrative and judiciary systems as well as in the media. As a result, the written form of 28 regional and local languages that were formerly used only in spoken form are now being used in primary education, in official communication, and in the media and the public spheres. Regional governments have developed their languages by creating scripts for them and are preparing teaching materials in them to meet the needs of the national curriculum. The Latin alphabet has been adopted for writing most of the country's languages while some have opted for the traditional Ethiopic script that Amharic uses. In the regional cities, regional languages have become visible in the public sphere: they are part of the linguistic landscape at federal and regional offices, businesses, shops, streets, a situation that is relatively new in the Ethiopian context. Moreover, English, the de facto official second language in Ethiopia, is in the equation of language use, alongside the national working language Amharic and the regional languages that have gained status through the policy of ethnic federalism.
A focus on tWo IMPortAnt rEgIons In EtHIoPIA: tIgrAy And oroMIA E xamining literacy practices in two regions that have been associated with linguistic and social struggles can provide insight into the impact of language policy and the ideologies associated with the languages. In the following, a general introduction to current linguistic practices in various domains in each region is presented. The two regions were chosen purposesively.
As the two largest languages of the country aside from Amharic, Oromo and Tigrinya occupied and still occupy a special place in the debate about language rights and language use in Ethiopia …. Oromos and Tigreans were in the forefront of the demand for the use of their own languages in all the apparatus of modern life in Ethiopia. Language rights were conceived as one of the first and most tangible facets of the recognition of other ethnicities' rights. (Appleyard and Orwin 2008: 277) Given the historical dominance of Amharic, one would thus expect a move towards divergence from that language in the assertion of various regional languages' newfound linguistic rights. tigray and tigrinya Tigray Regional State of Ethiopia, which is mostly inhabited by people of Tigrean (or Tigrayan) origin, is the northernmost of the nine autonomous regions of federal Ethiopia. Its current population is estimated at 4.3 million. Although there are major towns and urban areas, this federal region, with Mekele as the capital and administrative centre, is mainly composed of highland rural territory. Mekele, founded in the nineteenth century as a capital city by Emperor Yohannes IV, is a point on a major axis of urbanisation along the route from Ethiopia's capital city, Addis Ababa, to Asmara in Eritrea, and is located 650 kilometres north of Addis Ababa (Tamru 2007). Since being founded, Mekele has grown to be one of Ethiopia's principal economic centres. The city has greatly flourished and expanded under the current government with a significant population growth, making it the largest city in northern Ethiopia. According to Census 2007 (Addis Ababa: Central Statistical Agency, 2008) (the most recent available statistics), Mekele had an estimated total population of about 215,000, and a high population density. Given its significant growth and thriving commercial interests in the region, Mekele presents an excellent point of departure for investigating literacy practices in the linguistic landscape.
Tigrinya, the official language of Tigray, is spoken by nearly three million people in the federal region. Despite Tigrinya's dominance in Tigray, there are other minority languages spoken in the region, including Afar, Saho, Agew, Oromo and Kunama, all of which belong to other non-Semitic language families. According to the language policy of the region, Tigrinya serves as the official working language and the language of education, used as a medium of instruction from Grades 1 to Grade 8. Recently, Kunama (a Nilo-Saharan language) and Irob (also known as Saho, a Cushitic language) have been introduced as a subject in Grades 1 and 2 in the respective localities of the region. On the other hand, the role of Amharic in Tigray has been reduced and is currently only introduced to students as a subject from Grade 3 onwards. In addition the basic curriculum calls for English to be taught from Grade 1.
According to Ethnologue (Lewis 2009), the literacy rate in Tigrinya as an L1 is 1%-10% while in Tigrinya as an L2, it is 27%. No reliable up-to-date government statistics are available on literacy rates. Similar to Amharic, Tigrinya uses the Ethiopic script, also called Fidel. Tigrinya is also Eritrea's national language and since that country's independence from Ethiopia in 1993, the language has been developed making written material available. Tigrinya speakers in Ethiopia, however, do not have access to this material because of the political situation and also because cultural exchanges do not take place between the two countries. The introduction of the new language policy in Ethiopia, has had positive effects on the development of the language and literacy in the language has increased, relatively speaking, especially among the younger generation in Tigray.

oromia and oromo
Oromia is the largest of the nine federal regions of Ethiopia, in both size and population, covering a vast area of the south of the country. It has a population of about 24 million. This region covers most of the territory of the Oromo people, who had originally migrated into the area during the 16 th century and are now the largest single ethnic group in Ethiopia (Marcus 2002). Addis Ababa, the capital of Ethiopia, is located in this territory. It was considered the regional capital of Oromia until 2000 when the Ethiopian government moved the capital of Oromia to Adama, a city that is located along a major road in the region that connects the capital with other urban centers as well as with the port of Djibouti. This was a highly political move interpreted by some as an attempt to divorce the country's capital from the Oromia region and its people; however, the government insisted that the development of the Oromo language and culture, as provided in the new Constitution, would be best accomplished outside the capital of Addis Ababa. Adama, the new capital, officially reverted to its original Oromo language name in 2000, but its previous name, Nazreth, named after the Biblical Nazareth by Emperor Haile Selassie, is still widely used. In 2005, following the highly contentious national elections that resulted in the victory of the opposition to power in the city of Addis Ababa, the regional government of Oromia was moved back to Addis Ababa although Adama remains the cultural and economic hub of Oromia. According to Census 2007 (Addis Ababa: Central Statistical Agency, 2008), the population of Adama is about 220,000.
Oromo, which is also known as Afan Oromoo or Oromiffa(a), is the most widely spoken Cushitic language of the Afroasiatic Phylum. It is the official language of the federal region of Oromia, which is spoken as a first language by more than 25 million Oromo and neighboring peoples in Ethiopia and Kenya. Since 1991, under the new system of ethnic regions, Oromo has been introduced as a medium of instruction in elementary schools throughout the region (including areas where other ethnic groups live and speak their languages) and as a language of administration within the region. Oromo is written with a modified Latin alphabet called Qubee, which was formally adopted in 1991. Various reasons were given why the Ethiopic script could not be adopted, including claims that it could not capture Oromo phonemic structure. This, however, was refuted by the linguist Baye Yiman (1996). Pasch (2008) notes that the introduction of the Latin or Roman alphabet in Africa was the first attempt of mass alphabetisation on the Continent. However, Humery (2010) points out the pre-colonial African tradition of Ajami, in which non-Arabic languages were written in Arabic script. In Senegal, writers of Pular started using Roman script primarily to distance themselves from the dominant Ajami writing tradition for Wolof (Humery 2010). Similarly, the choice of the Latin alphabet by the Oromo people as the basis for Oromo literacy has indeed ideological roots and may be interpreted as an assertion of linguistic and cultural identity in contrast with the dominating Amhara and Ethiopic script. In the first six or seven years following the adoption of Qubee, it is believed that more texts were written in the Oromo language between 1991 and 1997 than in the previous 100 years. Literacy rates vary across the different Oromo-speaking areas from 1% to 15% (Lewis 2009). As with Tigrinya, no reliable up-to-date statistics are available for current literacy rates in Oromia.

MEtHodoLogy
According to Barton and Hamilton (2000), a social theory of literacy is methodologically grounded in linguistic ethnography. The linguistic landscape study of Mekele, Tigray (see Lanza and Woldemariam 2008 for details), was part of a larger ethnographically oriented study on language ideology and use in the federal region in which both authors participated, as they engaged in participant observation in the heart of the city.
In line with what may be considered the first wave of linguistic landscape studies, an important area of the city was chosen, the main shopping district that was precisely demarcated by certain streets and squares. A locally trained field assistant took digital photos of all tokens of environmental print found in the public domain including signs, names on buildings, advertisements, commercial shop signs and public signs on government buildings, amounting to a total of 376. The data were subsequently categorised as public or private, and according to the frequency of representation of specific languages and according to the visual presentation of the languages in the material, polarised as top and bottom. Furthermore, ethnographic interviews with randomly selected shop owners were conducted with the aid of local field assistants during which questions were asked concerning language choice on the shop signs. Tigrayan shop owners extolled the use of the regional language, as noted in their own shop signs while those whose shop sign was written in Amharic were themselves Amhara.
Based on their observation of language contact in the linguistic landscape of Tigray, the researchers decided to explore whether the situation was the same in other regions. The data from Adama were collected in a follow-up study undertaken by the second author. Before the data compilation, graduate students of linguistics at Addis Ababa University who came from that region were interviewed and consulted to elicit their intuitions concerning the matter. They confirmed having observed a similar phenomenon in Adama to that attested in Mekele, that is, language contact between Oromo and Amharic. Subsequently, two graduate students who used to reside and work in Adama were selected to do fieldwork and take pictures of various types of signage. About 100 pictures representing monolingual Oromo signs, bilingual signs involving Oromo and Amharic, and trilingual signs of Oromo-Amharic-English were collected. The overall linguistic landscape profile of Adama, though interesting, is not addressed in this study. For our purposes here, special attention is given to the examination of signs indicating language contact.
The linguistic landscape data were also complemented with data from randomly selected student textbooks and the accompanying teachers' manuals used in the two regions (see References). These were examined to check for language contact between Amharic and the target languages. These texts were deemed representative of other student textbooks by our research assistants. Graduate students, native in Tigrinya and Oromo respectively, were assigned to identify features they judged as not conventionally belonging to the language in which the textbooks had been written and comment on them.
These texts had been translated from Amharic into the respective regional language in accordance with the new language policy of ethnic federalism. The texts were first published in 1992 by the respective regional educational bureaus, with new editions issued later. During interviews with officers at these educational bureaux, we learnt that the textbooks had been prepared by a group of experts who were believed to have a good knowledge of the subject and who were first language users of the respective regional languages. The experts also had a good command of the Amharic language from which the materials were translated. In the Oromia region, where Oromo has been used as the medium of instruction up to Grade 8, textbooks for Grade 7 and 8 were translated from English as there is no comparable Amharic book at that level. In this study, we focused our attention on language contact between Amharic and the two regional languages in Ethiopia and thus have limited the scope of our inquiry to textbooks translated from Amharic for grades below Grade 7. Nonetheless we discovered during our study that English was claimed to play a role even in the texts meant for younger students.

LAnguAgE usE And LAnguAgE contAct In LItErAcy PrActIcEs
I n the following section, we address the literacy practices in the two regional capital cities. The linguistic landscape is construed as a 'discursively constructed space and consequently signs as 'frozen actions' by various actors' (Pietikäinen et al. 2011: 1); this analogy is applied to schoolbooks as well. Initially, we examine the texts that local shop owners write for their shop signs and then we examine educational textual materials, with language contact as the unit of analysis. The issue of language contact figures prominently in discussions of language change, which involves an innovation and the spread of that innovation. Whether features in a language can be considered contact-induced change or as having evolved through internal causes or developments within a language requires careful scrutiny not only from a synchronic perspective but also from a diachronic one, and there is debate in the literature (see Poplack and Levey 2009;Thomason 2011). Nonetheless in the Ethiopian context, an exploration into attested examples of purported language contact can provide the basis for further inquiry beyond structural description. Ethnographic observation has shown fIgurE 1 English, tigrinya and transliteration (Lanza and Woldemariam 2008) Transliteration: 'Photo central' English: 'Photo central' Transliteration: 'Video photo copy' Tigrinya: 'ID card laminating' that given the country's history and the abrupt change in language policy, people in fact perceive these instances of nonconventional language use as language contact between the previously dominant Amharic and the regional languages.

Multilingualism in the linguistic landscape in tigray and oromia
In Lanza and Woldemariam (2008), we investigated the distribution of languages in the signs of the main shopping area of Mekele, the capital of Tigray, where Tigrinya is the dominant regional language. Results indicated that the linguistic landscape exhibited three main languages in monolingual and bilingual signs: Amharic, Tigrinya, and English, and that the linguistic landscape in Mekele was indeed influenced translocally. As expected, signs of regional concern were in Tigrinya and signs with a national focus were in Amharic, with English variably co-occurring with each language. What was of particular interest was how through their local literacy practices, individual actors positioned themselves towards the official language policy and to multilingualism in their contribution to the linguistic landscape. Spolsky (2003) points out the need to distinguish between policy and practice in any language policy framework. An example of a bilingual sign that also includes transliteration from English is shown in figure 1.
In private signs, Amharic is still used despite the new policy of ethnic federalism although Tigrinya does appear more clearly in Mekele's linguistic landscape. This new language policy has promoted language emancipation from the perceived hegemony of Amharic. Hence with an ideology favoring ethnic regionalism, Amharic appears weakened, a view pervasive among many Ethiopians with some even proclaiming Amharic to be an endangered language. Nonetheless the influence of Amharic is still present in the linguistic landscape of Mekele.
The influence and dominance of Amharic is not only at the surface level of the signs in the linguistic landscape, LANZA AND WOLDEMARIAM © Lanza, Woldemariam and CMDR. 2014 fIgurE 2 A stationery shop in Mekele but also at the abstract grammatical level, as noted in many of the Tigrinya signs that employ Amharic word order (see Woldemariam and Lanza). A particularly relevant linguistic construction within which to examine word order is the noun phrase as shop signs invariably contain such constructions. Noun phrases in Amharic are right-headed while their equivalents in Tigrinya are left-headed. However, Tigrinya noun phrases following Amharic word order were attested in the linguistic landscape in Mekele. Such expressions, which are not considered acceptable Tigrinya forms, are believed to be the result of the influence of Amharic (Nega 2003). For example, various syntactic structures, several of which involve Amharic word order, were documented for the expression 'stationery shop'. In figure 2, we see such a shop sign in which the embedded noun phrase follows Amharic word order while the main noun phrase follows Tigrinya word order, although the lexical items are all from Tigrinya.
In Example 1, the structure is parsed indicating the respective word orders. such as the use of diphthongs (sequences of different vowels) and consonant clusters, were also noted. In some cases a particular Amharic functional element was included, as noted in figure 3 below.
In figure 3 we see the Amharic genitive marker ya occurring in both the Oromo as well as the Amharic version realised in different orthographies; the Oromo is parsed in example 2.
On the surface, it appears that Amharic in the public sphere remains dominant. This is evident not only in the wide occurrence of signs in Amharic, but also in its structure co-occurring in items from Tigrinya and Oromo in the respective regions.
English also figures in the linguistic landscape of Mekele and Adama. In postcolonial Africa, many have deplored the 'linguistic imperialism' of English, and critics of globalisation and its implications for trade in developing countries have resisted the use in the public sphere of languages such as English that symbolise power derived from colonialism. Ethiopia has no colonial past and thus an ideology of resistance to English is not evident in the public sphere. Hence looking at local agency in the appropriation and spread of English as a world language is relevant (see Blommaert 2010;Stroud and Mpendukana 2009). Through her fieldwork in Tanzania and Kenya, Higgins (2009) succeeds in undermining prevalent conceptions of English as a global language by drawing attention to how hybridity is crafted on English locally. In the Ethiopian context, the Ethiopian languages are predominantly placed at the top of bilingual signs involving English, as opposed to European mediascapes in which English is 'on top', as the preferred language for framing purposes (Androutsopoulos, 2012). In the Ethiopian context in fact there are many cases in which the use of English did not necessarily have a communicative function but rather a more emblematic or symbolic function as a marker of modernity. This is evident in figure 4 from Mekele, in which the word modern written in Latin script has been inserted in a sign written in Ethiopic script.
An English speaker without knowledge of the regional language would not understand the sign as referring to modern clothing as opposed to traditional dress. This sign indexes modernity through the very use of the Latin alphabet, recognised as English by local speakers, although they may not speak English themselves.
In sum, the general picture of literacy practices in the linguistic landscape of both Mekele and Adama shows the use not only of the regional language, a new development, but also the continued use of Amharic, in addition to English. What is of particular interest is the purported language contact between Ethiopian languages revealed in the linguistic landscape particularly at the abstract grammatical level.

Multilingualism in monolingual schoolbooks
The literacy practices described above are not just a phenomenon of the linguistic landscape. They also involve other textual production in the two regions under study. Signs in the linguistic landscape may be considered 'unregulated spaces' (Sebba 2009), as the shop owners independently make their signs. However, contrary to what may be expected in 'regulated spaces' where monolingual norms prevail, the same type of language contact occurs. This is indeed the case in such a regulated space as official media broadcasts. According to information from media consultants who work in Ethiopian broadcasting, the Amharic news is taken as the main source for the respective transmissions in Oromo and Tigrinya, as the regional programmers translate it into the respective languages of the program. The result is a language use that is perceived locally as somewhat stilted and that is associated with the context of media broadcasts. Another interesting case of textual production in 'regulated spaces' is the use of schoolbooks in Tigrinya and Oromo in the elementary schools in the respective regions. As noted above, literacy in regional languages is a fairly recent phenomenon and pedagogical materials in these languages have been developed since the language policy of ethnic federalism was first introduced. In the Tigrinya and Oromo textbooks we consulted, we found examples of expressions in both these languages that follow Amharic syntactic structure. This is particularly true of noun phrases, as was the case with the linguistic landscape. As mentioned above, in Amharic, noun phrases are right-headed, and in general, in both Tigrinya and Oromo, noun phrases are left-headed.
In the Oromo texts, we discovered examples of noun phrases that appear to follow Amharic word order. For example, while a unit of measurement conventionally occurs following a qualifying numeral as in digirii zeeroo, 'zero degree', the expression used is zeeroo digrii, that is, as in fIgurE 4 A tailor shop in Mekele (Lanza and Woldemariam 2008).
the Amharic zero digri. Similarly, while measurement expressions conventionally follow the pattern as in km 800, m167, etc., we found the reverse order: 800 km, 167 m, etc. (Grade 5 Social Science Book 2004: 19 and 22). This word order does not reflect current spoken Oromo in the region. Similarly, noun phrases like Baha Afrikaa 'East Africa' and Dhihaa Afrikaa 'West Africa' (30) would conventionally be written in the reverse order: Afrikaa Bahaa and Afrikaa Dhihaa. The presence of such expressions may be the result of the transposition of the Amharic structure in the translation process. In other words, in the process of recreating the text from the source language to the target language, the experts may have maintained some Amharic word order while substituting the lexical items with the target language. Notably, the Amharic word order is in fact the same as for English. When officers at the Educational Bureau of Oromia were asked why such a word order was used in the expressions above, they replied that it was to safeguard students from becoming confused when they encountered the word order in English in Grade 9, when English becomes the medium of instruction in Oromia. They claimed that following English word order for Oromo expressions would help facilitate the transition from Oromo to English at Grade 9. This observation deserves further inquiry. As previously noted, the basis for Oromo literacy has ideological roots opposed to the dominance of Amharic. Such a distancing from potential Amharic impact in the translation of the textbooks into Oromo may in fact be interpreted in light of this ideology.
The textbooks in Tigrinya demonstrated some Amharic features, with Tigrinya noun phrases following Amharic word order, similar to the findings in the linguistic landscape. The following examples are taken from Grade 3 Students' Tigrinya Textbook (1997: 62, 82, and 103) with the conventional form listed below each attested example.

5)
Attested structure ħriša kadrä agriculture cadre 'agricultural expert' Conventional structure kadrä ħriša cadre agriculture 'agricultural expert' Interestingly, Example 4 reflects a similar example attested in the linguistic landscape of Mekele, as demonstrated in Figure 2 above. Similar structural transpositions from Amharic were also encountered in the teacher's guide manuals (Grade 5 Teacher's Guide 2001:19).

7)
Attested structure ħɨto milkt question mark 'question mark' Conventional structure m ɨlkt ħɨto mark question 'question mark' In sum, while examples of language contact were attested in unregulated spaces, the linguistic landscape, where multilingual norms prevail, similar examples were found in regulated spaces in which there is an expectation that monolingual norms will prevail, indeed especially in light of the new language policy of ethnic federalism, which promotes the development of regional languages. This policy catapulted a change in literacy practices in the country and the data suggest that language contact, particularly involving Amharic with the regional language, serves as input for both informal and formal learning. The influence of English is also noteworthy. While these documented texts are not exhaustive data, they are nonetheless in line with ethnographic observations in the two regions.

dIscussIon And concLusIon
In previous centuries, literacy practices in Ethiopia were left to members of the elite and those belonging to religious communities of the Ethiopian Orthodox Church, an important institution in the country. The language of learning was Ge'ez, a now extinct Semitic language, and it was limited to the Bible and other religious works. Subsequently, Amharic assumed the role of Ge'ez especially in non-religious domains. Amharic's dominant role until 1991 created the opportunity for it to become the most developed literary language in the country. It became established as the language of literacy in all non-liturgical arenas until the introduction of the new language policy of ethnic federalism, which promoted the development and literacy of other regional languages. As illustrated in this study, even though regional languages such as Tigrinya and Oromo have assumed the position normally reserved for Amharic in their respective localities, people nevertheless still employ elements of Amharic structure in writing their own language. Such examples of language contact are not widely documented in the spoken forms of Tigrinya or Oromo, yet they appear in the linguistic landscape and other textual materials; and they are used in educational materials as well as in media broadcasting. As noted above, Ethiopians view the language used in media broadcasts as stilted. The sociolinguistics of multilingualism requires 'an approach to language from the vantage point of the social circulation of languages across spaces and different semiotic artifacts' (Stroud and Mpendukana 2009), such as in the various media involved in these case studies in Ethiopia. As previously noted, a pervasive language ideology in the country points to the general dominance of Amharic as an explanation for the language contact, in spite of the fact that Amharic is claimed to have a weaker role in Ethiopian society under the new language policy. The question then remains as to how to interpret these findings. One may be tempted to view the texts in question as 'poor' translations, often the interpretation given by Ethiopians. Yet the pattern of structural borrowing persists. One clear reason for such structural borrowing is the historical status of Amharic in the country and the intensive contact between the regional languages such as Tigrinya/Oromo, on the one hand, and Amharic on the other. Most of the speakers of regional languages are bilingual in Amharic, a situation that allows the languages to be in extensive and long-term contact. However, the type of language contact illustrated in this article is more restricted. It involves literacy practices, both in writing and in the transmission of written texts through media broadcasting. The shop owners belong to the generation for whom Amharic was the medium of instruction in schools and hence the key to literacy. Despite the regional acquisition of new language rights regarding written language in the public sphere, it seems that Tigrinya and Oromo speakers continue to perceive Amharic as the language of literacy even although they employ their own language in writing. This may be encouraged by the tacit acceptance of widespread language contact, and the fact that Amharic has traditionally been the language of literacy. Therefore, it can be said that this is evidence of the covert power of Amharic, despite its reduced role in current language policy.
Languages, however, are not agents; agency is a capacity of speakers. If we take the view of language as local practice, as opposed to the analyst's view of language contact between two separate 'reified' entities (see Makoni and Pennycook 2007), what we are witnessing is how speakers of the regional languages draw on their multilingual resources to create a new arena for language use -an arena that develops a new register for the regional language. Hence in the words of Makoni and Pennycook (2007), we as analysts need to 'disinvent' our conceptions of language as preset notions in regards to multilingualism, and rather 'acknowledge that languages are inherently hybrid, grammars are emergent and communication fluid' (Canagarajah 2007: 233). Indeed research on code-switching counteracts the monolingual bias in dealing with language contact (see Auer 2007). Multilingualism should not be seen as merely a collection of 'languages' that a speaker controls, but rather as a complex of specific semiotic resources, some of which belong to a conventionally defined 'language', while others belong to another 'language'. (Blommaert 2010:102) Such a functionalist view of linguistic resources is certainly empowering to speakers (Matras 2009). These Tigrayan and Oromo speaker-writers can therefore be perceived as active agents in adapting their multilingual competence to new linguistic practices. Indeed the language contact, which has been witnessed in literacy practices since the introduction of the new policy of ethnic federalism, appears to have an indexical value signaling a new register of language use, hence a process of potential enregisterment (Agha 2005) that deserves further attention. Agha (2005: 38) defines processes of enregisterment as 'processes whereby distinct forms of speech come to be socially recognised (or enregistered) as indexical of speaker attributes by a population of language users', that is, the process by which they become 'ideologically linked with social identities' (Johnstone 2011: 657). Such enregisterment has been proposed for a language contact situation by Babel (2011) in a study of speakers of Andean Spanish where she noted that some features borrowed or calqued from Quechua appeared in their speech. Babel demonstrates that 'while some Quechua contact features in Spanish have been linked to informal contexts associated with Quechua and Quechua speakers through the process of enregisterment, other features bypassed this process, retaining Quechua pragmatic content that links them to a formal register' (56). Indeed more detailed studies of spoken language use in the contact situations reported on here are needed before any specific conclusions can be drawn on the possibility of enregisterment. Nonetheless contact features such as those documented here in the written language are recognised as indexing literacy practices pursuant to the new language policy promoting regional languages.
The linguistic landscape and the textbooks investigated in this article were created by a generation of speakers educated when the national working language Amharic was the language of instruction and literacy. A new generation of Ethiopians is coming of age, having been educated and socialised under the new policy of ethnic federalism during the 1990s, with texts similar to those under study in this article. To what extent this new generation will maintain these literacy practices is the subject of future research. Indeed these speakerwriters may well become active agents of language change not only within more written language use but also in spoken discourse. More data on other potential language contact features in both spoken and written discourse are needed to explore these issues within the Ethiopian context. While structural change that is induced by contact is usually attributed to social processes, Matras (2009: 310) argues for the role of multilingual speakers as agents of language change: While it is obvious that isolated individuals are not in a position to introduce changes that will shape a language's diachrony, it is also crystal clear that no societal process can lead to language change unless it prompts individuals to innovate their own speech. The shop owners and sign designers as well as the textbook writers are all potential agents of language change. The examination of local literacy practices in both regulated and unregulated spaces in two federal regions in Ethiopia provides insight into the role of both individual agency as well as social processes in language contact phenomena. Further work on new emerging literacies, for example, in text messaging and online chat forums, can provide further insight into individual speakers' agency in the use of their linguistic repertoires, including language contact. This exploratory investigation into language contact in literacy practices has focused on a country considered marginal, or peripheral, from a global perspective yet issues concerning centre and periphery are also at play. The cities in focus in each region are indeed centres within the margin. Examining language contact shows how multilingualism and literacy practices are enacted across trajectories of centre and periphery. Speakers/writers of regional languages draw on resources, both their linguistic and literacy skills, from the language of the centre in the country, and the traditional language of literacy in creating new literacies in their formerly marginal languages within the regional context. The marginality of the regional capitals and their centrality within that marginality may be considered as an explanation for the language contact phenomena studied in this article. The analysis of the similarity of language contact patterns of the two remote spaces, particularly in the respective linguistic landscapes, implies an approach to linguistic landscapes in terms of a sociolinguistics of multilingual mobility rather than linguistic localisation, where one would need to look at how multiple encodings of a discourse are transfigured across signage, contexts and languages. (Stroud and Mpendukana 2009: 381) Indeed the use of the regional languages, the national working language, and English invoke various discourses of cultural and linguistic identity, tradition and modernity, and ultimately social mobility through literacy. Mazrui (2004) argues that while English has linked Africa to the world, it has paradoxically both enriched and reduced many indigenous African languages by marginalising them. Such a view is prevalent in the discourse of endangered languages. In the Ethiopian context, however, English does not carry the colonial baggage it carries elsewhere in Africa and Asia. As noted particularly in the educational domain, English is introduced in the curriculum before Amharic, the language that had previously dominated and marginalised other Ethiopian languages. With the new policy of ethnic federalism, some formerly marginal languages in Ethiopia have now gained central roles in the transition to literacy. As Shohamy (2006: 110) points out so clearly, the presence or absence of languages 'sends direct and indirect messages with regard to the centrality versus the marginality of certain languages in society'. Indeed the true marginal languages today are those that have not been documented in the linguistic landscape and the educational materials -the numerous minority languages in the country that are marginalised at the margin without sponsored literacy.
EndnotE 1 This work was partly supported by the Research Council of Norway through its Centres of Excellence funding scheme, project number 223265.