As illustrated in Section 2, numerous studies exist on different multilingual regions and their linguistic landscapes. Galicia, as illustrated by the demographic data in Section 1, is indeed a multilingual region. However, as will be discussed in this section, work on the LL of Vigo remains scarce even though the city is the largest (by population) in the autonomous community. Vigo is also well known outside the region, for example, for its industrial and tourist port. One of the most recent works, also encompassing Vigo’s LL, has been conducted by Suárez Caramés (2025), who compares it with the LL of Lugo. She analyses two districts in each city with different socioeconomic backgrounds. One of her key findings, evident in other contexts in Galicia as well, as discussed in this section, is that Spanish dominates on private signage, followed by English and then Galician. Signage by authorities, i.e., top-down signage, follows the recommendation of keeping official communication in Galician. Another key finding is that, especially in the context of advertising, English takes over symbolic functions, which comes as a consequence of globalization (Suárez Caramés 2025: 282).

Rodríguez Barcia and Ramallo (2015: 150) discuss the relevance of the Galician language as an expression of (local) identity in the context of graffiti in Vigo. Galician is the most frequently used language in their dataset for graffiti. The authors argue that this dominance of Galician along with the often socio-political messages communicated through the graffiti implies a link between language and social demands. This observation is supported by the fact that the vast majority of graffiti in Galician are indeed addressing a “local” or “universal” audience (compared to those in Spanish, of which roughly a quarter topicalizes “egocentric” topics) and that barely any graffiti in Galician were found that do not concern “social or public” matters (while only half of the Spanish-language graffiti discuss such themes) (Rodríguez Barcia & Ramallo 2015: 142-143). The authors consequently highlight that the practice of expression through language can be perceived as an act of strengthening (linguistic and political) identity. This observation adds a further dimension to the study of the LL of Vigo, as the choice of language must also be interpreted as an act of identity and – potentially – political ideology. In this context, Zas Varela and Prego Vázquez’s (2016: 22) analysis of the LL in public urban spaces in Arteixo (A Coruña, Galicia) shows that especially graffiti is a means to emplace languages in the LL that are otherwise invisible. The authors also stress that English, a language not part of the local linguistic repertoire, occupies central spaces in public signage overall, while the co-official language, Galician, is limited to institutional signage, in line with other studies. However, they note that the presence of English is often fragmentary (i.e., one-word passages) and usually implemented in combination with Spanish, with the latter conveying the major information (Zas Varela and Prego Vázquez 2016: 13). Overall, Zas Varela and Prego Vázquez (2016: 22-23) note that local varieties of both native (especially Galician) and migrant groups are displaced to peripheral areas, while international languages (especially English) and glocal linguistic resources claim central spaces of the square analyzed, also leading to displacement of local voices and further fueling the commodification of internationally renowned languages.

Table 1

	< 10.000 inhabitants	10.000-20.000 inhabitants	20.000-50.000 inhabitants	> 50.000 inhabitants
Galician	54.85%	39.41%	28.37%	17.25%
Spanish	13.63%	24.00%	37.65%	51.14%
Both	28.13%	32.05%	30.45%	27.90%
Other situations	3.39%	4.53%	3.53%	3.71%

Adapted from Instituto Galego de Estatística. (2024). Encuesta estructural a los hogares: Uso del gallego. Xunta de Galicia. Retrieved January 10, 2025, from https://www.ige.gal

Moving towards other cities in Galicia, Dunlevy (2012), for example, compares the status of Galician and Spanish in rural and urban areas in the north of the region. A comparison of the cities of A Coruña and Cee has revealed that, even on official signs, Galician is much less common than Spanish in the metropolis of A Coruña, despite this not complying with local language policies. This was not the case in rural Cee, where most signage followed official policies. In both cases, Spanish is the language most commonly used on private signs, but the difference in frequency of use between the two languages was considerably smaller in the rural area than in urban A Coruña (Dunlevy 2012). If one compares these findings with the language use of the inhabitants of these cities, various overlaps are visible. Table 1 compares the mother tongue(s) of the Galician population by the number of inhabitants of their cities of origin.

While Cee, as a small town, can be placed on the left side of Table 1, the city of A Coruña is located on the right side of the spectrum of population numbers. Regarding private signage, there are comparatively more signs in Galician in Cee than in the metropolis. About official signage, Dunlevy (2012: 3) notes that the historical context of Spain's language policy must be considered here. While Galician was banned under the dictatorship of Francisco Franco (1939-1975), which had a major effect on the region's metropolises, the language continued to be used in many rural areas of Galicia. As a result, many villages in Galicia served as safe havens for the regional language, reflected in public signage, which was changed from Galician to Spanish in some cases in the larger cities during the dictatorship.

A study of the linguistic landscape of Pontevedra, a city with about 80,000 inhabitants, has revealed a notable difference in the choice of language for signage depending on the type of sign (for further details on types of signs, see Section 2.2). The data collected as part of the study were divided into administrative ("sinais administrativos") and commercial ("sinais comerciais") signs. While 96% of administrative signs are in Galician, thus following the requirement that official communication, especially by the authorities, should be carried out in the regional language, only 11.2% of commercial monolingual signs are in Galician. In addition, most of the 635 signs examined are monolingual (75.3%), of which the majority (80%) are in Spanish (López Docampo 2011: 20-23).

Regueira, López Docampo & Wellings (2013: 51) confirm these findings in a comparative study of the cities of Pontevedra and Santiago de Compostela. The latter city is known as the destination of the Way of St. James, a pilgrimage route for many believers. This is why the city attracts a large number of tourists from around the world. Consequently, the study has revealed a slightly larger proportion of signs containing English in Santiago de Compostela than in Pontevedra. Overall, however, Regueira, López Docampo & Wellings (2013: 51) have shown that for both cities the proportion of Spanish signs is larger than the combined proportion of all signs in other languages. In addition, they note the dominance of Galician on institutional signs compared to those installed by private actors, which is interpreted as a consequence of local language policy (Regueira, López Docampo & Wellings 2013: 57-58). Mosquera Castro and Wellings (2014: 179-181) observe the same trend for the region of Galicia as a whole. López Docampo (2011) finds that in Santiago de Compostela only 17% of monolingual private signs are in Galician. This finding runs counter to the population's preferences, as Ramallo and Rei Doval (1997: 325) found in a study of language attitudes. Their results show that more than 83% of participants consider the use of Galician for commercial purposes to be either “good” or “very good”.

The present study aims to build on these earlier studies in two ways. First of all, the city of Vigo, the largest in Galicia in terms of population, should be included in the research discourse to provide a more detailed picture of the linguistic landscapes in the region (cf. RQ1). In addition, a survey of Vigo will provide further insights and material for comparison with the contexts introduced earlier in this section. Secondly, the study at hand compares two shopping streets in the city of Vigo that cater to different clienteles. On the one hand, Rúa do Príncipe is the city's main shopping street, geared towards locals and tourists alike. On the other hand, Rúa Vía Norte is a street opposite a large shopping center and is home mainly to service businesses on the ground floors of its buildings. On the upper floors, numerous apartments are located. Rúa Vía Norte, therefore, does not attract tourists the same way Rúa do Príncipe does. Furthermore, in Rúa do Príncipe, various offices and service industries are located, which are not to be found in Rúa Vía Norte. All in all, Rúa Vía Norte tends to be less tourist-oriented due to the focus of its services. Therefore, another aim of the paper is to examine whether the commercial orientation of Rúa do Príncipe affects its linguistic landscape and, if so, how. The methodology for addressing the two RQs of the study will be introduced in Section 3. Since our analysis will factor in a number of predictors potentially influencing language choice and distribution in the linguistic landscape of Vigo (cf. RQ2), we will first introduce earlier research and findings on the most important factors, i.e., type of sign, type of multilingual writing, and considerations on the influence of proper nouns on language choice and distribution.

As discussed in Section 2.1, signs can be grouped by characteristics and classified into types. However, before we turn towards these and the resulting types of signs in more detail, we will discuss and determine what constitutes a sign in our study. In this respect, the traditionally employed definition of the linguistic sign goes back to Backhaus (2007: 66), who defines the sign as “any piece of written text within a spatially definable frame”. However, answering this question is by no means trivial, as Gorter & Cenoz (2024: 114) aptly note (see also Blackwood 2015; Gorter 2013): “[T]he unit of analysis can be hard to define. Decisions about the unit of analysis are important because they concern crucial methodological issues which imply selection processes and can allow for comparability (or not) between studies.”

In addition, Piritidis (2014) has shown that how we define our unit of analysis may directly influence our study's findings. A shop window, for example, can be treated as one single unit of analysis which consists of several different parts (e.g., the advertisement of the goods to be sold as such, but also opening hours, stickers, posters, etc.) since it constitutes “a spatially definable frame” (Backhaus 2007: 66). On the other hand, these different parts can be treated and analyzed as signs in themselves since they are “spatially definable”, too. A shop window may therefore consist of a number of signs and thus may contain more than one unit of analysis. Gorter & Cenoz (2024: 389) discuss further objects that may constitute units for analysis. These include (1) three-dimensional ones, some of which are spatially separated from one another but are inseparably connected in context such as lettering on buildings (cf. Savela 2018); (2) moving objects, including T-shirts worn by passers-by (cf. Järlehed 2019); or (3) digital and virtual signs, such as display boards or QR codes. In the study at hand, we follow the more fine-grained approach, for a number of reasons (cf. Section 3), but in particular since a narrow analysis of types of signs, to be introduced in the following, would not be possible when, for example, treating the overall shop window as the unit of analysis.

For a detailed analysis of signs, different types have been identified, and earlier studies have shown that these can have an important influence on language use and preferences. In this respect, Scollon and Scollon (2003: 167) and Ziegler et al., (2018: 79) propose the following categories, which we later use for our analysis (cf. Section 3.2):

i. commercial: Names of shops and all signs that refer to services or products. Also, signs displayed in shop windows, unless they can be explicitly assigned to another category.

In addition to the two official languages, Spanish and Galician, previous studies of commercially oriented areas in Galicia have shown that various other languages can be expected to play a role on signs in such areas (cf. López Docampo 2011). For the German context, for example, Ronan (2022: 42-43) has shown that Italian is used to express culinary competence, French is employed on signs to evoke the association of the advertised products with "elegance", and English is used to create a sense of internationality. Since the city of Vigo is located near the Atlantic Ocean and is a tourist port, we would also expect to find English, the lingua franca of tourism, as part of the local linguistic landscape, in particular in Rúa do Príncipe. Assuming the use of two or more languages on one sign for the context investigated, we, therefore, have to reflect on different types of multilingual arrangement of languages on these signs, as identified in earlier literature. These different types of multilingual writing refer to the exact relationship between several languages on a sign. Reh (2004: 8-15), for example, and, expanding on her classification, López Docampo (2011: 19) identify the following categories:

i. duplicating: The text is available in more than one language and the content is identical, i.e., it is simply duplicated.

Another aspect of the analysis of signs concerns the use of proper nouns. As Gorter & Cenoz (2024: 357) note, these may include “a personal name, a brand name or a product name, also a building name or any other type of name”. Assigning these names to languages can be challenging, if not problematic. An example of this can be found in a study by Tufi and Blackwood (2010), which shows that, for example, brand names are assigned to certain languages depending on the recipient. In this study, the brand name Diesel, for example, was assigned to Italian by Italian speakers but not to a specific language by French speakers. In addition, the word exists as a proper noun in other languages, such as German (Cenoz and Gorter 2024: 358), too. Related to this problem, proper nouns, such as brand names, seldom constitute instances of real language use that provide information about deliberate language choices and preferences. To counteract this problem, Buschfeld, Weihs &Ronan (2024: 313) suggest to label signs accordingly, for example, indicating whether the sign displays a proper noun only or a proper noun plus additional text. For indicating the use of proper nouns with unclear or ambiguous linguistic origin, they suggest coding the data by means of the label “brand name unclear”. A less complex alternative would be to exclude signs that display proper nouns only. This is done for the brand names and proper nouns in the study at hand (cf. Section 3 for details).

The data were collected along approximately 260 meters of the Vigo shopping streets Rúa do Príncipe and Rúa Vía Norte (cf. Section 2.1), as illustrated in Figures 1 and 2. Both figures mark the starting point in green, the endpoint in red, and the path taken during data collection in yellow. It should be noted that the area of investigation of the street Rúa do Príncipe includes both sides of the street and the objects in between, such as benches or decorations. For Rúa Vía Norte, only one side of the street was selected, as the other side is mainly covered by the roof of a shopping center, which is not comparable to the other stretches examined.

Figure 1 Figure 1. Area of investigation: Rúa do Príncipe Google LLC. 2025. Google Earth Pro (7.3.6.10201)

The present study is based on a total of 271 photos, 159 taken in Rúa do Príncipe and 112 in Rúa Vía Norte. From these photographs, we extracted 251 signs according to the definition of a sign provided in Section 2.2 (138 in Rúa do Príncipe; 113 in Rúa Vía Norte). As mentioned in Section 2.4, signs displaying only a proper noun were excluded, which accounts for the number of photos taken exceeding the number of signs analyzed. The same signs that were visible several times were also counted several times, as this increases the visibility of the language(s) used just as much as different signs with the same language distribution. The data were collected on December 14, 2024, by the first author of the study, using a mobile phone camera.

Figure 2 Figure 2. Area of investigation: Rúa Vía Norte Google LLC. 2025. Google Earth Pro (7.3.6.10201)

Since our study aims to investigate language use on signs immediately visible to passers-by in the two streets introduced, and thus to determine their linguistic landscapes, signs were analyzed that were placed on the ground floor of buildings directly adjacent to the streets and were thus immediately visible to passers-by. Those signs placed in higher stories may not directly catch the eye of passers-by and so they were not considered. Signs are considered visible if they can be seen from a position at 90° to the building wall. Figure 3 illustrates this consideration.

Figure 3 Figure 3. Visibility of signs as a passer-by

This restriction was made because some of the building entrances were set back and signs attached to the side of the sidewalls are only recognizable if pedestrians walked past them at a right angle (see red markings in Figure 3; the signs displayed in the green marked parts of buildings were therefore included in the data set). When it comes to the question of what to treat as a sign, i.e., an independent unit for analysis, we follow an approach similar to that of Gorter & Cenoz (2024: 389) and Savela (2018). Carriers of information constituting a coherent and semantically closed unit, i.e., which display contents that immediately relate to each other and are semantically and/or structurally inseparable, even if not necessarily physically connected (cf. Figure 3), were counted as individual signs. In addition, signs had to be attached to a carrier physically connected to the study area (cf. Backhaus 2007: 67). In the example in Figure 4, the lettering “Se podes soñalo, podes facelo” counts as one sign. The objects behind the shop window are not considered, as they are not physically connected to the carrier (here: the shop window).

Figure 4 Figure 4. Lettering on a shop window

In addition to these criteria, it must be noted that the present dataset does not include any digital or non-permanent signs. Non-permanent signs are those that are regularly moved. These include, for example, vehicles, garbage lying on the ground, café chairs that visitors can move, and signs from street artists or vendors who have to change location at short intervals due to legal regulations. We decided to exclude them from the analysis because these objects are highly dependent on the point in time at which the investigation is conducted. Still, we decided to include the Christmas decorations in the pedestrian zone of Rúa do Príncipe, since they are illuminated from November and are normally installed much earlier, making them visible for several months. It can therefore be assumed that the presence of these objects is comparable in terms of their permanence to the seasonal advertisements of local retailers. It must be noted that, since 2018, the aim of the famous Christmas celebrations in Vigo has been to attract national and international visitors, and therefore the use of foreign languages, especially English, is prominent in these decorative installations. On the other hand, expressions such as “Merry Christmas” are frequently used on Christmas decorations in a wide range of non-English-speaking countries.

Furthermore, only signs that contained at least one complete word were considered; signs that displayed images only were excluded from the analysis, as our study focuses on questions of language use.

After data collection, each sign was assigned a four-digit number and was coded according to various characteristics (henceforth, variables), as introduced and described in Table 2, along with the abbreviations used for the statistical analysis in Section 4.3.

Most of these variables, their realizations, and coding labels (their abbreviations indicated in brackets) are straightforward and do not need further explanation (e.g., criterion “Location”) or have been introduced as part of the theoretical framework in Section 2 (criteria “sType” or “mType”). However, the criteria “nLng” and “Languages”, as well as “PrN” and “Author” require some further explanation, which we will provide in the following:

While the criterion “nLng” appears to be straightforward at first sight and language on a sign can usually be identified without any issue for speakers of this language, Anglicisms (and other loan words) can still make this task a challenge. As the question whether words such as the English term “parking” are already sufficiently integrated into the Spanish language to be considered Spanish cannot be answered straight away, we decided to treat all Anglicisms as English to avoid inconsistency in these decisions. As discussed in Section 2.4, similar difficulties can arise when distinguishing between Spanish and Galician on signs, as both languages share a lot of vocabulary. Consequently, if no clear-cut distinction could be made, both languages were considered present, and the type of multilingualism was considered “identical”.

Table 2

Variable	Levels
Location	Rúa do Príncipe (RP); Rúa Vía Norte (RVN)
Number of languages (nLng)	1; 2; 3; >3
Languages (binary with y for „present“ and n for „not present“)	Spanish (ESP); Galician (GAL); English (ENG)
Type of sign (sType)	infrastructural (i); regulatory (r); commercial (c); transgressive (t); commemorative (cm); other (o)
Type of multilingual writing (mType)	duplicating (d); fragmentary (f); overlapping (ov); complementary (cp); identical (id); monolingual (0)
Proper nouns (PrN)	Proper noun plus text (pn+); no Proper noun (0)
Author	official (of); private (pr)

Regarding the criterion “PrN”, other than in Buschfeld, Weihs & Ronan (2024), proper nouns were excluded from analysis if they stood on their own, for a number of reasons, most importantly since the linguistic origin of proper nouns, i.e. which language they should be coded as, can often not be determined reliably. This criterion therefore distinguishes between “no proper noun”, coded by 0, and “proper noun plus text”, coded by pn+. In the latter case, only the additional text on the sign was coded for all criteria mentioned in Table 2. It must be noted that the presence of the proper noun on the sign did not influence any of these criteria (e.g., the proper noun itself cannot make the sign multilingual as it is disregarded for the coding). However, the variable whether a sign contains a proper noun (i.e. “pn+”) or not (i.e. “0”) may have an influence on the text part of the sign and was therefore used as a potential predictor in the statistical analysis presented in Section 4.3.

In addition, regarding the criterion “Languages”, many places in Galicia have both a Spanish and a distinct Galician name (for example, Plaza and Calle (sp.) compared to Praza and Rúa (gal.) for square and street). Consequently, place names, such as those of squares or streets, were included in the coding, as their languages are straightforward to determine and differ between Spanish and Galician, the two languages expected to be the most frequent options in our context. Therefore, when analyzing the languages on a street sign, for example, the sign does not only refer to a place that has a fixed name in one of the languages, but whomsoever put up the sign must also decide which language(s) to use. Language use on street signs was thus treated as productive text and not as proper nouns (cf. the discussion in Section 2.4).

In addition, we provide some further details on the coding of authorship. In our study, official signs are those installed exclusively by the city administration. For this purpose, signs with the inscription "Concello de Vigo", the name of the city administration, its logo, and the names of its departments (for example "Concellería de Cascos Históricos") were coded as "official". This does not include, for example, posters advertising an event supported by the city administration. Although the name and logo of the city administration are usually present on such signs too, the city administration is usually not the author but simply listed as a supporter of the event. Such posters and all other signs were assigned the code "private".

For data analysis, all signs were coded according to the categories introduced in Table 2 and the full set of information on each sign was entered into a Microsoft Excel spreadsheet. Following data coding, various analytical methods were used to examine the research questions (cf. Section 1) from different perspectives. First of all, the frequencies of languages on signs and the influence of the different variables identified earlier in this section on their realization were measured quantitatively and modeled statistically, using both descriptive and inferential statistical methods. In addition, particularly interesting findings were analyzed qualitatively, i.e., highlighted and discussed as individual examples that still illustrate more general, frequent patterns.

In the following Section (Section 4), we begin with a descriptive illustration of our results, i.e., tables and bar charts, to first evaluate which languages dominate the linguistic landscape of Vigo in terms of relative frequencies (cf. RQ 1). We here focus on the influence of type of sign and authorship on language choice. In addition, an inferential statistical analysis was carried out to determine which of the factors from the list of characteristics above (our independent variables, cf. Table 2) has a significant influence on language use on signs (our dependent/target variable) (cf. RQ 2). To this end, we employed conditional inference trees (also known as ctrees; Hothorn, Hornik & Zeileis 2006) in RStudio (Version 2024.12.0+467). Ctrees are often used as an inferential statistical tool in linguistic studies (Gries 2020; Hothorn, Hornik & Zeileis 2006; Tagliamonte & Baayen 2012) because they can handle smaller data sets and are comparatively easy to interpret (cf. Section 4.3). In general, ctrees employ statistical tests for inference and automatically identify and include only those decisions that significantly improve the correct prediction of, in our case, language choice on signs in Vigo.

Of the 251 signs analyzed in the dataset, 170 contain Spanish either in isolation or in combination with other languages, making it the most frequently used language. Galician is used on 91 signs, followed by 54 signs containing English. Other languages used once each are Portuguese and Italian. However, concerning the first three languages, we found differences between the different types of signs they occur on. These findings are illustrated in Figure 5. Four of the six types of signs, i.e., infrastructural (69.7%), commercial (70.5%), transgressive (43.1%), and other (75%), feature Spanish most frequently; on the other two types of signs, Spanish is the second most frequent language. English was found in only four of the six categories, most prominently on commercial (32%) and transgressive (26.2%) signs and never on regulatory signs or signs belonging to the category “other”.

Figure 5 Figure 5. Language distribution by types of signs including Spanish, Galician and English

Galician holds the largest shares for regulatory and commemorative signs with 65.2% and 69.2%, respectively. In the present dataset, these are especially no-parking signs in Rúa Vía Norte and memorial plaques, exclusively present in Rúa do Príncipe. It is striking that all of the memorial plaques (9) are inscribed with Concello de Vigo, the city administration (or in one case the Department of Historical Buildings), and thus have been put up by an official authority. The same applies to no-parking signs such as the one illustrated in Figure 6. However, regulatory signs put up by private authors predominantly feature Spanish and point out, for example, a ban on bringing pets into a hotel lobby (Figure 7).

Figure 6 Figure 6. No parking sign in Rúa Vía Norte

Figure 7 Figure 7. Sign prohibiting dogs in a hotel lobby

This can be treated as the first evidence suggesting that authorship is indeed an important factor in the linguistic landscape of Vigo. Zooming in on this aspect of the analysis, we identified 27 signs that can be classified as “official” (cf. Section 3.2). These include all nine commemorative plaques, six infrastructural and 12 regulatory signs. Among them are no-parking signs and road signs. Four of the 27 signs of official authorship are multilingual. Two of those are no-parking signs in the identical multilingual format; they can be received in Spanish and Galician. Two others include English, Spanish, and Galician. Those are commemorative plaques (cf. Figure 8) that translate the text but display the heading and historical context in Spanish and Galician, which are identical in both cases. Consequently, the type of multilingual writing would be duplicative for the text part, but the entire sign needs to be classified as a fragmentary type, as the heading is not provided in English. Regarding the languages on official signs, it can be stated that all 27 signs contain Galician. 13 of them are located in Rúa do Príncipe, 14 in Rúa Vía Norte. In the following Section (4.2), we focus on these aspects related to multilingualism.

Figure 8 Figure 8. Memorial plaque in Rúa do Príncipe

Regarding the combinations of languages found on the signs examined, it should first be noted that 192 of the 251 signs are monolingual. Table 3 shows the distribution of signs by the number of languages (nLng) used.

Table 3

nLng	1	2	3	>3
Amount	192 (76.5%)	54 (21.5%)	4 (1.6%)	1 (0.4%)

Of the 192 monolingual signs, most are in Spanish (59.4%), followed by 26.6% in Galician, 13.5% in English, and one sign in Portuguese (0.5%). If we change perspective and look at the proportion of monolingual signs among all signs of a language, it is noticeable that signs in Spanish not only make up the majority of monolingual signs, but, conversely, most of the Spanish signs are monolingual. Although this distribution also applies to signs that contain Galician, the ratio is lower. For signs that contain English, the distribution is almost balanced, with a slight tendency in favor of multilingual signs. Table 4 presents the detailed numbers and percentages for the distribution of monolingual and multilingual signs by language.

Table 4

	Spanish	Galician	English	Total
Monolingual	114 (67.1%)	51 (56.0%)	26 (48.1%)	192 (76.5%)1
Multilingual	56 (32.9%)	40 (44.0%)	28 (51.9%)	59 (23.5%)

¹

Including one monolingual sign in Portuguese.

In addition, the question arises as to how languages relate to each other. Therefore, the signs were coded according to types of multilingualism. In this respect, the 59 signs identified as multilingual can be divided into four of the five types of multilingual writing described in Section 2.3; overlapping multilingualism was not found. Complementary multilingualism was the most common type observed for 29 signs, and thus about half of the multilingual signs require knowledge of two or more languages to understand all the information. 24 of these 29 signs contain English. An example of such a sign with complementary multilingual writing is shown in Figure 9. The other multilingual signs are divided into 25 signs that contain Spanish and Galician. Since these two languages are very similar, if not identical in our coding of multilingualism, multilingualism may have been less intended by the authors than in the other types of multilingual signage. In addition, our analysis identified three signs of duplicating multilingual writing and 2 further signs of fragmentary multilingual writing. These results are illustrated in Table 5.

Table 5

mType	Amount
Duplicating	3 (5.1%)
Fragmentary	2 (3.4%)
Overlapping	0 (0%)
Complementary	29 (53.7%)
Identical	25 (42.4%)

Furthermore, the combination of Spanish and English on multilingual signs offers interesting insights into contact patterns between the two. For example, signs in which the English word “parking” is integrated into a Spanish matrix sentence were found (cf. Figure 10). Figure 11 shows a sign that uses English to describe the type of business advertised. Sometimes the specifying information on such signs is in Spanish, but in many cases it is communicated in English. The content on this sign, i.e., the business's opening hours, is not in English but in Spanish and Galician, which allows interpreting English as a symbol of the professionalism and competence of the business owner. The local official languages are then preferred for the informative function of the sign.

Figure 9 Figure 9. Sign using complementary multilingual writing

Figure 10 Figure 10. Sign at the entrance to a parking garage in Rúa Vía Norte

Figure 11 Figure 11. Sign at the entrance to a parking garage in Rúa Vía Norte

Other factors that influence the type of multilingualism on a sign are the type of sign and whether it contains a proper noun. Concerning the latter, those that contain a proper noun and additional text are more likely to feature multilingual text (pn+: 29.5% multilingual) than those that do not contain a proper noun (0: 19.2% multilingual). Regarding the type of sign, commercial and infrastructural signs are most likely to feature multilingual text (cf. Figure 12). Bringing together these observations, it should be noted that 21 of the 40 multilingual signs in these two categories also contain a proper noun in addition to their multilingual text. Figure 12 shows the distribution of multilingualism across all types of signs.

Figure 12 Figure 12. Type of multilingualism by type of sign

Since the descriptive statistical analysis suggests that different characteristics of the signs (e.g., authorship, type of sign, type of multilingual writing, etc.) have an influence on the display of languages on public signage in Vigo, the questions arise as to how exactly these relate to each other and whether their influence is truly significant. Using inferential statistical analyses (cf. Section 3.2), we shed light on these questions in the following and identify several significant predictors.

The ctree in Figure 13, for example, illustrates the significant predictors for the appearance of Galician on a sign. The model presented has a balanced accuracy, i.e., the accuracy calculated on both the small and large classes (“no Galician present” and “Galician present”), of 95.618%. The first and most important factor (Node 1) is authorship. As noted in Section 4.2, all 27 signs by official authors use Galician, as confirmed in Node 2. The dark gray parts of the boxes in the terminal nodes show the predictions for Galician and the light gray parts indicate the predictions for “no Galician included in the sign”, abbreviated by y (yes) and n (no). For signs by private authors, as displayed on the right-hand side of the ctree, additional factors play a significant role in language use. Node 3 splits the signs according to type of multilingual writing (cf. Table 2 for an overview of abbreviations). Node 11 predicts that all signs with duplicative or identical multilingual writing also contain Galician. However, if the sign is monolingual or of the “complementary multilingual writing” type, the next significant predictor is the presence of Spanish on the sign (Node 4). If Spanish is absent and no English is on the sign (Node 5), almost all signs (except the monolingual Portuguese sign) contain Galician (Node 6). If English or Spanish are included on the sign, the number of signs in Galician is much lower (Node 7, 9, 10). It is noteworthy that, even on complementary multilingual signs with Spanish (Node 10), the majority of signs do not include Galician, and Spanish is more likely to be combined with other languages, primarily English. In our dataset, these are primarily instances of Anglicisms, i.e., terms that have entered the Spanish language in a way that even people with low or no proficiency in English would understand their meaning. These are either fixed expressions such as “Merry Christmas” (cf. Figure 9) or integrated into Spanish matrix sentences or statements on signs (cf. Figure 10). This suggests a general influence of English on Spanish (as found in most languages worldwide) and language contact in Vigo (as discussed in Section 4.2 in more detail).

Figure 13 Figure 13. Prediction of Galician

Figure 14 Figure 14. Prediction of Spanish featuring all variables

Looking at the predictors for Spanish on signs, Figure 14 shows that the presence or absence of Galician (Node 1) is the most significant predictor. Yet, most of the predictions made by this ctree are no-brainers (balanced accuracy: 98.406%), for example, Node 11 predicts that Spanish is not found on signs that contain Galician, but only this one language (Node 7).

Consequently, we decided to model a further ctree for predicting Spanish, excluding the number of languages (nLng) and the type of multilingual writing (mType) as variables. The latter was also excluded, as the type of multilingual writing is modeled similarly to the number of languages, with, for example 0 (mType) equaling 1 (nLng). The balanced accuracy of this ctree is 88.048%. It is shown in Figure 15.

Figure 15 Figure 15. Prediction of Spanish with reduced predictors

Although the presence or absence of Galician remains the most significant predictor in the model (Node 1), the presence of Spanish, especially when Galician is present on a sign, is influenced by several other significant factors. First, authorship (Node 7) has again proven important here: If it is official and Galician is on the sign, the presence of Spanish is unlikely (< 20%; Node 8). This confirms the findings from our descriptive quantitative analysis (cf. Section 4.1). In contrast, for signs by private authors featuring Galician, the probability of Spanish on an infrastructural sign is very high (Node 13). In addition, for other types of signs, the presence of a proper noun plus text (pn +) plays a vital role (Node 12). As illustrated on the left-hand side of the ctree, the type of sign has a highly significant influence on the presence of Spanish when Galician is absent, and English is present (Node 4). While infrastructural signs feature Spanish in addition to English in all of the observed cases (Node 6), this is not the case for commercial and transgressive signs (Node 5). In these cases, it is more likely that Spanish is not used alongside English.

In the following Section (5), we interpret and place the findings in the context of the research questions and the earlier studies presented in Section 2.