Abstract
The paper investigates the placement of subject and finite verb in topicalized, i.e. non-subject initial declarative main clauses in North American Danish. European Danish adheres to the V2-rule and thus requires inversion, while North American Danish allows for non-inversion, i.e. [X]SV word order. Based on a sample of approx. 1700 tokens of topicalized declarative clauses produced by 64 speakers, we observe a general stability of V2 in North American Danish. In order to explain the instances of non-V2, we employ both linguistic and sociolinguistic factors.
1 Introduction
The non-inversion of subject and verb in non-subject initial declarative main clauses (topicalized clauses) is a recognized feature in Germanic heritage languages in North America. This paper adds to the field of research a corpus-based study of this phenomenon in American Danish. Standard Danish, a typical V2 language, strictly adheres to the so-called V2 rule, whereby the finite verb is in second position and the preceding field is filled by exactly one constituent (XVS). Accordingly, Standard Danish non-subject initial declarative main clauses require reverse order of subject and finite verb, i.e. XVS, in order to adhere to the V2 constraint.
1
Examples (1a) and (1b) below are thus acceptable in Danish. American Danish seems to follow the V2-rule in non-subject declarative clauses as well, but simultaneously allows for an English-like ‘V3 syntax’ in which a topicalized non-subject constituent and the subject may occur before the finite verb (XSV). Example (1c) below is ungrammatical in Standard Danish but nevertheless present in American Danish.
2
Our aim is to give a descriptive account of the variation in subject-verb order in American Danish in non-subject initial declarative clauses and to define the linguistic, demographic, and sociolinguistic factors that facilitate syntactic variation in this construction.
Our focal point of analysis is the deviation from European Danish (i.e. XSV word order), but our analyses also produce results for the occurrence of the opposite, i.e. the European Danish word order XVS. We thus emphasize both divergence from and accordance with European Danish word order under the conditions of migration.
The American Danish immigrant group has, generally speaking, been quite open to influences from the surrounding majority society. This implies that we must take into account the influence from English as a subject-first language. However, other factors seem to have an impact as well.
Because this study is committed to an empirical and descriptive approach, our labelling of the feature that we investigate will be purely descriptive during the analyses, naming it XVS or XSV. We do so to avoid reference as to whether either should be considered occurrences of V2 or V3 by classifying them within specific frameworks.
The paper is structured as follows: Section 2 contains a review of previous research on word order in topicalized main clauses in other West or North Germanic Heritage varieties spoken in North America in order to generate hypotheses that can be checked against our data. The dataset for our study is detailed in section 3, while section 4 gives a brief account of Danish migration to North America. Section 5 specifies our speaker sample. This is followed by detailed analyses of our dataset in section 6 and a discussion in section 7. Section 8 contains concluding remarks.
2 Review of Previous Research
Because of the genetic and typological closeness of other varieties of North and West Germanic languages spoken in North America, these are adequate means of comparison for the syntactical development of American Danish. Furthermore, these varieties exist under the same conditions as American Danish inasmuch as they are spoken by immigrants or their descendants as heritage languages. As such, they are under constant pressure by the dominant majority language English in the assimilation-oriented American society.
Though the variation in verb placement in non-subject initial declaratives is definitely recognized as a feature in which Germanic varieties may undergo changes, published findings on the occurrence of XSV word order are scarce. In general, the German varieties in North America show remarkable stability in their SOV and V2 characteristics (cf. the various German varieties discussed in Salmons (1993), Louden (1994, and p.c.) on Pennsylvania Dutch, Boas (2009) on Texas German, Nützel and Salmons (2011) on various German speaking language islands in the American Midwest and Hopp and Putnam (2015) on Moundrigde Schweitzer German). An exception to this general stability emerges in Sewell’s study of syntactic variation in Wisconsin German narratives, in which she notes instances of ADV+SV word order in three out of eight speakers and attributes this to English influence. The ADV+SV word order occurs specifically in clauses starting with hier ‘here’, dann/denn ‘then’ and jetzt/wella ‘now’ (Sewell, 2015: 242–244).
Van Marle (2001: 94ff, referring to Gehring, 1973) mentions the remodeling of American Low Dutch word order in topicalized main clauses in accordance with English, i.e., XSV word order, and relates the change to contact with English. However, referring to the paucity of reliable American Low Dutch sources, the author himself advises caution as it is unclear to what extent the English counterparts of the Dutch constructions have taken over, to what extent the Dutch patterns remain in use, and what kinds of speakers use them.
Newman (2015) discusses V2 loss in word order in topicalized main clauses in the Yiddish of New York Hasidim. She focuses on consecutive word order in which the V2 rule has the discursive function of indicating subsequence and consequence of the first clause onto the following clause. She concludes that Hasidic New York Yiddish is moving towards English’s subject-first word order, although the change is far from completed. Newman also finds that time adverbials introduce the SV word order more frequently than other adverbials (Newman, 2015: 194).
Turning to North American Icelandic, a North Germanic variety in North America, Arnbjörnsdóttir (2006, 2015) states that word order in this variety generally is resilient to change. Although she gives an example of a non-subject initial declarative main clause with V3 word order (Arnbjörnsdóttir, 2006: 109), she does not discuss whether and to what extent this is representative for the syntactical development of North American Icelandic.
Recently, several papers have provided us with analyses of word order in American Norwegian and American Swedish. Change in main declarative word order in non-subject initial main clauses is observed in both languages but is reported to be a rare and recent phenomenon. Larsson et al. (2012: 280f) find no evidence of non-V2 word order in Hasselmo’s (1974) early reports of American Swedish, while Eide and Hjelde (2015: 24f) conclude for American Norwegian that “the V2 rule was intact and very robust at Haugen’s time” referring to the early recordings by Haugen in the 1940s (cf. Haugen, 1953). Changes, however, occur in recent American Swedish and American Norwegian. Discussing Heritage American Norwegian in the 1990s, Hjelde (2001: 225) states that, whereas “European Norwegian only allows one constituent to the left of the finite verb […] American Norwegian, like English, lacks this so-called verb-second constraint” (Hjelde, 2001: 225). Hjelde considers this to be a contact-induced phenomenon.
The lack of the verb-second constraint in Heritage Norwegian is further explored by Eide and Hjelde (2015: 14), who discuss whether a receding finiteness distinction might effect “the V2 machinery in these contact varieties of Norwegian,” i.e. in Heritage American Norwegian collected in the 1990s and 2010. In the data from the 1990s, they find that Norwegian word order “is still robust,” V2 violations “are rather scarce,” and there are no “profound changes in the language structure at this point in time” (Eide and Hjelde, 2015: 28f). In their latest dataset from 2010, there is still no “significant tendency toward the disintegration of V2,” but the authors now observe significant individual variation, with one informant in particular, a 3rd-generation speaker of American Norwegian, showing “a very strong tendency towards disintegration of the V2 structure” (Eide and Hjelde, 2015: 32). The authors note that this is particularly the case for heavier fronted topics and when the subject is a pronoun (Eide and Hjelde, 2015: 40). They see this (incipient) change as a mixing of “competing systems” and, referring to Matras (2009: 151, 237), they state that this is “exactly what one would expect in a language contact situation” (Eide and Hjelde, 2015: 46).
Larsson and Johannessen (2015a, b) combine the analysis of American Norwegian and American Swedish. They focus mainly on word order in dependent clauses related to the notions of language attrition and incomplete acquisition, but briefly mention word order in main declarative clauses. A clear finding is that the homeland word order system is intact in the speech of the original immigrants. This is in line with the findings of Eide and Hjelde (2015), in part because they refer to the same historical datasets (Haugen’s and Hasselmo’s recordings).
Larsson and Johannessen agree with Eide and Hjelde (2015) that word order change is observed only in recent recordings of Heritage American Norwegian and Swedish. However, their interpretation of the word order variation differs from that of Eide and Hjelde in that they see the occurrences of XSV word order in main declarative main clauses as signs of attrition rather than an effect of English influence (Larsson and Johannessen, 2015a: 172f, 183; Larsson and Johannessen, 2015b: 255f). They state that only a few tokens of XSV word order occur and that these can be attributed to “less competent and confident” speakers, who speak slowly and with difficulty, show morphological instability, and produce simpler sentences (Larsson and Johannessen, 2015b: 255f). Overall, they consider the variation in word order to be “performance errors rather than as changes in the underlying grammatical system” due to the speakers’ old age and limited recent productive use of the Heritage language (Larsson and Johannessen, 2015b: 257; 2015a: 174). They consider the influence of English and the special role that fronted elements may have in information structure, though without coming to definite conclusions. 3
This short review of previous research shows a quite muddled picture with regard to variation and change of syntax in non-subject initial clauses in different Heritage Germanic varieties: The German heritage varieties 4 seem to be very resilient to change. The Mainland Scandinavian varieties show change only in recent recordings and only produced by heritage speakers, not immigrant speakers. Syntactic change is recognized for North American Icelandic and early American Low Dutch, but with no possibility of assessing the degree of variation. New York Hasidic Yiddish seems to be on the road towards XSV word order. This review of previous research on syntactical change, especially with regard to word order in declarative clauses with an initial non-subject constituent, provides us with several hypotheses that we will pursue in section 6.
3 Data
The data sample that forms the basis of this study is part of a larger corpus consisting of informal speech collected in the 1960s and 1970s by two Danish linguists, I. Kjær and M. Baumann Larsen. Altogether, they succeeded in recording approximately 200 Danish Americans (original immigrants and their descendants) in the USA and Canada. 5 The data in our sample consists of 54 rather short (15–25 minutes) and structured interviews by one of the researchers with one or two informants at a time. The structured approach of the interviews sometimes impedes free and unguided speech, and the shortness of the interviews allows little time for hesitant speakers to recollect words and phrases. However, the strategy of recurrent questions by the interviewers provides us with quite consistent information throughout the data sample concerning, for instance, dialectal background, visits to Denmark, participation in Danish-American “communities of practice” (Eckert, 2006), and language use (Danish or English) by the informants’ children and grandchildren. In addition, the collectors have carefully noted relevant information on all speakers (names, year of birth, place of origin, emigration year, place of residence).
Nevertheless, the data sample for our study has come about uncontrolled, at least by us: 46 interviews existed as transcripts dating from the late 1980s and were thus easy to process and analyze with regard to word order. We are not able to determine why exactly these particular interviews were chosen for transcription out of the whole dataset, as the collectors’ decision-making process is not documented, and the speaker sample itself suggests no obvious decision criteria. In addition to the older transcriptions, we have added eight newly digitalized and transcribed interviews to the study. At the time of analysis (summer 2015), the data sample amounted to approximately 103,000 words.
4 Danish Migration to North America 1900–1930
Although the Danish migration did not empty entire regions in the same way as in the other Mainland Scandinavian countries, Denmark provided its share to the European mass emigration to North America around 1900. Between 1900 and 1930, 140,000 Danes left for North America and remained there (Grøngaard Jeppesen, 2005: 265ff.). By 1920, 189,000 Danish-born Danes lived in the USA, and together with the first US-born generation, this amounts to 468,000 people (Grøngaard Jeppesen, 2005: 264). This meant that a considerable proportion of all Danes lived in the USA (6% Danish-born and 15% if the first US-born generation is included), but it was nevertheless a rather small number compared to other European migrant groups. Between 1900 and 1930, the migrants from Denmark were mainly young, unmarried people with a rural background, with a high percentage (55–65% annually) of single young men (Hvidt, 1971: 188ff). This in turn led to extensive intermarriage with other groups, mainly of Northern European ancestry (Grøngaard Jeppesen, 2005: 282ff).
The preferred area of cohesive settlements for the Danish group was the Midwestern states, but in total, only around half of the Danish immigrant group chose these states as a new home. Some of the Danish migrants settled in the rural Midwest, close to the other Mainland Scandinavian groups, but many preferred the big Midwestern cities, especially Chicago and Minneapolis (Grøngaard Jeppesen, 2011: 19ff.). In general, the Danish immigrants and their descendants did not settle close to fellow countrymen. They were geographically mobile, moving according to job opportunities and with few ties to the company of other Danish Americans or the existence of Danish institutions such as Danish churches. Due to their social and geographical independence from fellow Danes and Danish cultural items, the group did very well socioeconomically (Grøngaard Jeppesen, 2011: 105f, 179), acquired English rapidly (Grøngaard Jeppesen, 2005: 283), and apparently gave up Danish quickly. 6 In turn, this means that, although there have been Danish enclaves (e.g. in Racine County, Wisconsin, and some districts of Chicago) where Danish language, culture, and traditions were upheld in formal or informal networks or “communities of practice” (Eckert, 2006), such enclaves were rare compared with those of other migrant groups.
5 Speaker Sample
Small migrant groups in a non-colonial setting (i.e. groups with no intention or even a remote possibility of linguistically influencing the host country’s future) are very sensitive to certain extra-linguistic factors such as settlement patterns, internal network structures, and recognition of the group within the host country (by the group itself and by others). Furthermore, sociopsychological factors like motivation for emigration, attitudes towards the host country (assimilation versus segregation), perception of the usefulness of the migrant language, and the migrant language’s status as a core value in the group’s cultural value system have a great impact on the linguistic outcome of smaller migrant group, especially in interplay with the host country’s ideology with regard to minorities (for different ideologies, see Bourhis 2001: 9ff). This suggests that demographic, macro-sociolinguistic, and socio-psychological factors must be taken into account when analyzing the linguistic outcome of each specific migrant group. Grouping them together as heritage speakers (in the sense of Rothman 2009 and Polinsky 2011; cf. footnote 4) might fail to account for differences in outcome.
Triggered by our experience with the demographic, sociolinguistic, and linguistic heterogeneity of the Danish migrant group (heterogeneous with regard to settlement patterns, marriage patterns, degrees of connectedness with other Danish Americans, and accordingly patterns of language loss and language maintenance), we believe that a detailed description of the speaker sample is of particular importance for analyzing linguistic outcome. This claim seems to hold for migrant groups in general because these non-indigenous groups rely neither on a historically defined territory nor on an autochthonous speaker group nor any other self-evident relationship between territory, nationality, and language as speakers of national language of nation states. Migrant groups are often far from being closed entities but instead represent a heterogeneous sample of speakers from a certain home country. As a result, a sample can vary greatly with regard to age, gender, place of origin, social class, and linguistic/dialectal background, and the migration process will cause further social diversification. Research in heritage languages has rightly pointed out that the sociolinguistic variables of the home country cannot be self-evidently transferred to the migrant group after emigration but must instead be established in their own rights, as must the speakers’ baseline language (Polinsky and Kagan, 2007; Aalberse and Muysken, 2013).
Our speaker sample contains 64 speakers: 49 original immigrants born in Denmark (22 women, 27 men), 14 first-generation US-born Danish Americans (8 women, 7 men), and 1 second-generation US-born Danish American woman. In the following, we will distinguish ‘immigrant speakers’ and ‘heritage speakers’ in accordance with Putnam and Sanchéz (2013: 478–479) and Montrul (2009: 241). Heritage speakers are here defined as all speakers born in America or speakers born in Denmark but migrated (with their families) before “the period of later language development that takes place during the pre-school and school years (4–13 years)” when “many aspects of grammar may not reach full development and remain incompletely acquired” (Montrul 2009: 241). We define immigrant speakers as all speakers who were born in Denmark and migrated after “the period of later language development.”
About half of the immigrant speakers (n = 29) come from Jutland, the Western continental part of Denmark, the remaining speakers from all over Denmark, with no special concentration in other regions. The age of the informants at the time of recording covers a great range, with the youngest speaker being 40 years old and the oldest being 97. The majority of the immigrants (n = 48) migrated 1900–1930, and their length of residence in North America at the time of recording ranges from 16 to 86 years (average = 57). In some respects, this implies that our sample might contain (fossilized) historical speech. However, with regard to word order in declarative clauses with topicalized elements, today’s Danish does not differ from the historic Danish dialects or regiolects between 1850 and 1900, which the parents of the heritage language speakers spoke and which constituted the childhood input for the immigrant speakers. This means that, at least for this particular feature, we do not need to consider intra-language variation and can compare the American Danish word order in declarative main clauses with a non-subject initial constituent with today’s Standard Danish language.
Our speaker sample shows a quite diffuse residence pattern, with little geographical concentration of speakers at the time of the recording. This is shown in Table 1 below.
Since the Danish immigrants and their descendants were characterized by a high degree of geographical mobility and an overall tendency towards non-cohesive settlements (cf. section 5), this sample might actually represent the group quite well. Furthermore, geographical placement at the time of recording seems to be of less significance with regard to language use and maintenance than (reported) participation in (Danish-speaking) Danish American communities of practice. In a previous qualitative (and as yet unpublished) study, we linked extra-linguistic factors such as participation in Danish networks, language practices in the family, and contact with Denmark to individual language loss and language maintenance. 7 However, we found that a majority of speakers (n = 82, partial overlap with the dataset for the present study) showed little participation in Danish American networks and generally little nostalgia for Denmark or engagement with Danish language or culture.
As mentioned earlier, we can conclude that the present study’s speaker sample shows no self-evident selection criteria on the part of the original interviewers. The speakers form a meta-group based on a few constituting features: The speakers are migrants from Denmark or their descendants, they spoke enough Danish to have been eligible for the collectors, and they were willing to participate in interviews.
6 Analysis
The sentence-initial elements which we have taken into account in our analysis of word order in American Danish are of the following constituent types: Adverbials of various semantic (temporal, locative, modal, etc.) and syntactic categories (adverbs, prepositional phrases, subclauses, or combinations of these), objects, constructions with a left-dislocated constituent that is re-established by an anaphoric particle (usually så), and what we have termed adverbial-connectors. These are så ‘then, so’, nu ‘now’, and da ‘when, then’, often following the conjunctions og ‘and’ or men ‘but’. In Standard Danish, så, nu, and da are defined as either adverbials or conjunctions. In standard Danish, these words can only be attributed to word classes by referring to word order. SV word order define them as conjunctions (cf. example 2 below), while VS word order define mark them as adverbs (cf. example 3 below).
Since American Danish shows variation in word order in non-subject initial main clauses, word order cannot be used as an indicator of word class assignment for these words. The other factor that Standard Danish grammar writing uses for the definition of word class are the semantics of så, nu and da: As conjunctions they indicate consecutive meaning, as adverbs they indicate temporal meaning. However, in the genre of chronological biographical narratives, these semantics merge and are hard to tell apart. Consequently, we consider these elements as a separate analytical category. By doing this, we are able to investigate whether SV word order in clauses starting with an adverbial-connector (a shared option by the languages in contact) might be ‘overused’ by the speakers of American Danish.
In the following analyses, we single out factors that have a bearing on word order in main declarative clauses with fronted elements. We thus focus on divergence from European Danish word order, implying convergence towards English word order (XSV), as well as the contexts in which European Danish/Scandinavian word order (XVS) remains intact. Given the possibility of expanding the analysis with further data, we see this study as a means of approaching the intricate relationship between word order on the one hand and sociolinguistic and linguistic factors on the other. This gives us the opportunity to generate hypotheses and results that can be taken up and tested against more data in the future.
6.1 General Distribution
Our dataset includes 1707 declarative main clauses with a non-subject initial constituent. 154 (9%) of these display XSV word order while 1553 (91%) show XVS, the European Danish word order. We can therefore conclude that XSV word order has not become predominant in the context of main declarative clauses with fronted elements, yet we cannot exclude the 154 tokens as occasional processing errors by dysfluent speakers. 8
The 154 XSV clauses are not spread equally across the 64 speakers in our sample: 37 (59%) speakers produce XSV clauses while the rest do not. If a speaker produces XSV clauses, s/he typically produces one or two of these constructions (median = 2; mean = 4.2), but the intra-speaker variation is great, varying between 100% to 1.4%. For example, speaker OBO has 1 XSV construction out of 118 main clauses (8%) with a fronted non-subject constituent while speaker HAO has 6 out of 8 (75%). As a consequence of this skewness, we refrain from testing the distributional patterns with, for examples, statistical measures of independency and stick to analyses on the basis of percentages figures.
We have two outliers in the data: Female HEF with 19 XSV tokens out of 39 relevant clauses and male ADJ, with 63 XSV tokens out of 91. We have decided to exclude the extreme outlier ADJ in the present analysis, since his extreme and idiosyncratic use of the XSV word order may have a bias on the analysis.
6.2 Sociolinguistic Factors
As our speaker sample contains both immigrant and heritage speakers, this provides us with the opportunity to relate our findings on word order to speaker type. Larsson and Johannessen (2015a, b) observe a pronounced difference between immigrant speakers and heritage speakers of American Swedish and American Norwegian with regard to word order: In their datasets, only heritage speakers produce the English word order XSV (cf. section 2). This intra-group difference does not hold for our speaker sample, as shown in Table 2.
As can be seen in table 2, XSV word order is distributed rather equally across speaker types, 4%–6%. The slightly larger share of XSV that the immigrant speakers show does not give reasons to assume that this background factor has a significant influence. Furthermore, both the ‘minor’ outlier HEF (included) and the ‘extreme’ outlier ADJ (not included) are immigrant speakers, and some heritage speakers do not produce XSV word order at all. We therefore conclude that the intra-group difference noted in American Norwegian and American Swedish does not apply to American Danish.
Further, we looked at the potential correlation of word order with other sociolinguistic and demographic factors such as 1) gender, 2) place of residence at the time of recording as an indication of participation in Danish-speaking environment, 3) time spent in the USA for the migrant speakers as a measure of exposure to English, and 4) year of birth for the heritage speakers as a measure of possible exposure to XSV word order in the baseline heritage variety of Danish that forms the input of their language acquisition (assuming a growing rate in the heritage variety as compared to European Danish). The distribution of the two word order patterns did not for any of these factors indicate a potential correlation with these speaker demographics.
6.3 Linguistic Factors
As the sociolinguistic and demographic factors do not correlate with the observed variation in word order in declarative main clauses, linguistic factors remain as possible causes. In the following analyses, we follow the paths of related work on other Heritage Germanic varieties in North America (reviewed in section 2) as well as our own agenda and ideas.
We have considered 1) the type of the fronted constituent and (referring to Newman, 2015) the difference between time adverbials and other adverbials as fronted constituents (section 6.3.1), 3) the quantitative scope of the fronted constituent (referring to Eide and Hjelde, 2015, section 6.3.2), 4) the type of the subject (pronominal versus nominal subjects; referring to Eide and Hjelde, 2015, section 6.3.3), 5) the occurrence of English elements as (part of the) fronted constituents (section 6.3.4), and 6) the occurrence of left-dislocated constituent resumed by an anaphoric particle before the finite verb (section 6.3.5).
6.3.1 Fronted Element and Constituent Type
In order to gain a clearer picture of which linguistic factors might influence the occurrence of XSV and XVS, respectively, we have grouped the fronted constituent into three categories: Fronted adverbials (single words, phrasal units, or subclauses; ADV), adverbial connectors (see above; ADV-CON), and other non-subject constituents (e.g. objects, subject predicates, or dislocated arguments for prepositions; OTH). Table 3 shows the distribution of XSV and XVS across these three categories.
From the table it appears that the XSV word order occurs more often with a fronted adverbial (ADV) than with an adverbial-connector or another constituent as ADV topics account for 60% of all tokens with SV word order. The table also shows that XSV word order is not particularly rare with an ADV-CON, 33 out of 91 XSV occurrences, but that this word order only occurs in 4.1% of all ADV-CON sentences. This implies that we see no ‘overuse’ (seen from a standard language perspective) of XSV word order in clauses with an initial ADV-CON. We also see that OTH constituents before the subject-verb-cluster do not seem to have a noticeable effect on SV-order. We conclude from this that the XSV word order is more likely to occur with a fronted adverbial constituent than with others, but with some caution since the data do not allow for statistical testing.
Although fronted adverbials seem to trigger SV-word order more often than other non-subject constituents do, Table 4 (below) shows that there is no noticeable difference whether the preposed adverbial is a time adverbial or not (non-temporal adverbials (cause, place, circumstance, etc.; see Newman 2015) are grouped together as ‘other adverbials’) as these types of adverbials co-occur to an almost identical degree with XSV word order).
6.3.2 The Quantitative Scope of the Fronted Non-Subject Constituent
Eide and Hjelde (2015: 40) hypothesize that word order in main declarative clauses with fronted non-subject constituents is influenced by the quantitative scope of the constituent, i.e. that ‘heavier’ fronted constituents favor SV word order. In this study, we have measured the ‘weight’ of the fronted constituent by the number of words before the subject-verb cluster, but included also conjunctions.
9
Example (4) shows a short fronted constituent (an object, with conjunction, underlined) of 2 words, example (5) a longer fronted constituent of 5 words (underlined), combined with an initial adverb (both examples from our dataset).
There is a strong correlation between number of words and the type of fronted constituent. The mean number of words of fronted objects and dislocated arguments for prepositions before the subject-verb cluster is 1.7 words, for subject-predicates it is 1.2 words, for ADV-CON it is 1.6 words, and for ADV it is 4.5 words. We therefore decided to exclude OTH topics and ADV-CON constituents, keeping only the ADV constituents, like in (6) and (7), in the analysis.
By only considering fronted adverbial constituents we find that longer adverbial constituents tend to be followed by VS; with 10 or more words in the constituent, we see exclusively XVS word order. SV word order occurs most often after short adverbial constituents (36 out of 55 instances with XSV word order consist of only 1–4 words). However, the average difference in number of words before the two word order patterns is not pronounced (4.1 versus 4.5). Further, if we exclude a quite untypical adverbial constituent consisting of 21 words from the analysis, the ratio changes drastically: Fronted ADV constituents followed by SV word order will then be one word shorter than fronted ADV constituents followed by VS word order (3.8 words followed by SV versus 4.8 followed by VS). We are therefore tempted to conclude that, at least for adverbial constituents, the quantitative scope of the fronted constituent either makes no difference at all for the word order, or, that longer non-subject constituents actually favor European Danish VS word order. Both conclusions go against the hypothesis put forward by Eide and Hjelde (2015: 40) for American Norwegian.
6.3.3 The Type of Subject: Pronominal vs. Nominal Subjects
Following another hypothesis advanced by Eide and Hjelde (2015), we have focused on type of subjects in the main clause following a fronted non-subject constituent, dividing the subjects into pronominal and nominal subjects. In this analysis, we have excluded 123 tokens with the dummy subjects der ‘there’ and det ‘it’, as they do not have referential function or function as proper arguments in the sentence (Hansen and Heltoft, 2011: 1210–1233) as well of instances with det ‘it’ where it is not possible to decide whether it unambiguously functions as an anaphoric pronominal subject or a dummy subject.
Table 5 clearly shows that that type of subject does not correlate with a particular word order pattern in American Danish. This finding also points in another direction than hypothesized by (Eide and Hjelde, 2015: 40).
6.3.4 Occurrence of English Elements within the Fronted Constituent
A factor that sticks out as typical for XSV word order is the occurrence of an English lexical element prior to the subject-verb cluster. These English elements are mostly conjunctions or connectors such as and, then, so, and but, which occur either singly as the initial word in the sentence or as a part of a larger Danish constituent, but they may also be content words, as shown examples (6) and (7), respectively, from our American Danish dataset.
Table 6 shows the distribution of word order in relation to the occurrence of an English element. We observe a very clear preference for XSV word order when clauses are introduced by an English lexical element.
Bearing in mind that fronted ADV constituents also showed a noticeable tendency to occur with XSV word order, it is relevant to examine whether these are single effects or a confluence of both factors, i.e. the English element and ADV constituent type. In other words: Do the ADV constituents occur more often than other constituent types in the fronted position that contain English elements and are they in this case succeeded by SV word order? In order to answer this question, we need to take a closer look at the effect of constituent type by introducing the initial element as a constant factor. However, we exclude the ADV-CON from the analyses in order to avoid testing the same factor twice as the ADV-CON might actually be an English element. Table 7 shows the combined distribution of these factors.
In table 7 we see that when clauses do not have an initial English element, there are slightly more occurrences of SV word order when the fronted non-subject constituent is an adverbial. This supports the observation that adverbials favor XSV word order as shown in Table 4. But what about an initial English element’s possible isolated effect on word order? The best way to test this would be to look at word order in clauses with an initial English element and a fronted constituent other than ADV, but this occurs only 5 times in the dataset, and only with VS word order. But if we consider adverbial constituents alone, we see that they co-occur with SV word order to a much larger degree when they contain or are preceded by an initial English element (7.1% vs 65%). This indicates that the occurrence of an English element strengthens the effect of an ADV topic to be followed by SV word order.
6.3.5 Left-Dislocation Constructions with an Anaphoric Particle
In the preceding analyses, we have focused on factors that we suspected would co-occur with SV word order in declarative clauses with a fronted non-subject constituent. The following analysis sets out to analyze a syntactic construction in which the European Danish word order XVS appears to be robust throughout. This construction contains a left-dislocated constituent that is taken up by an anaphoric co-referent (in Danish, mostly pronominal det ‘that, it’, så ‘so, then’; but also da ‘then’, and adverbial der ‘then’), followed by a finite verb + subject (cf. Heltoft and Hansen, 2011: 1827ff.). The type of fronted constituent in these contexts varies from subordinate clauses to a variety of adverbials, as shown in examples (8)–(10) from American Danish, with the anaphoric particle in bold.
Table 8 shows the construction in the American Danish data in relation to the occurrence of SV and VS word order.
In the table we see that both topic constructions, without or with anaphoric particle, heavily favor VS word order, but also that a construction with a left-dislocated constituent with an anaphoric particle only very reluctantly (1.7%, 5 tokens) is followed the SV word order. We see this as a tendency towards a dispreference of the English word order pattern when the fronted constituent is resumed by an anaphoric element.
There is a certain correlation in the data between the effect of the anaphoric particle construction towards VS word order and the effect of English elements since an English element only rarely occurs within the left-dislocated constituent (4 tokens, 1.4%). This indicates that the syntactic construction with a left-dislocated constituent, followed by an anaphoric particle (which does not have an equivalent in English), meshes poorly with clause-initial English elements.
Also the effect towards SV word order that fronted adverbials seem to have (Table 5), is impeded by a construction with a left-dislocated constituent and an anaphoric particle. This is shown in Table 9.
Table 9 shows that when an adverbial constituent is fronted, as a left-dislocated constituent, and referred to by an anaphoric particle, we very rarely see XSV word order (1.8%, 5 tokens). When we do not have a construction with a left-dislocated constituent and an anaphoric particle, XSV word order is more likely to occur. We see this as an indication of a preference for XVS word order (European Danish word order) for clauses with left-dislocated constituents referred to by an anaphoric particle. Eide and Hjelde (2015) mention this construction, which is similar in Norwegian and Danish, in their analysis of Heritage American Norwegian. They notice that the construction with a co-referent marker generally keeps the European Norwegian word order VS, just as it does in our data. Yet, in their newer data, they find some occurrences of SV word order (Eide and Hjelde, 2015: 83f.) as does Johannessen (2015).
6.4 Main Findings
We observe a high degree of syntactical persistence in our sample of Danish American speech where European Danish XVS word order is predominant. However, 5–9% (depending on the inclusion of the outlier speaker ADJ) of the sample’s approximately 1700 tokens of declarative main clauses with a fronted non-subject constituent display the word order XSV, as in English. The distribution of XSV and XVS word order does not correlate with speaker generation (immigrant versus US-born heritage speakers), nor do any of the other sociolinguistic and demographic factors that we have investigated (gender, place of residence at time of recording, time spent in the USA for migrant speakers, year of birth for heritage speakers). The sample shows a great deal of inter- as well as intraspeaker variation.
While no sociolinguistic factors seem to correlate with the variability in word order, a number of linguistic factors appear to do so. First, fronted adverbial constituents seem to favor SV word order more than adverbial-connectors and other types of fronted constituents. Second, the occurrence of an English element within the fronted constituent correlates with SV word order. Third, there is either no correlation between word order and topic length (measured in number of words), or a weak correlation between longer fronted constituents and VS word order. Fourth, European Danish XVS word order is noticeably stable within a construction with an anaphoric particle following the fronted constituent. Linguistic factors that do not show any effect on word order in this data sample are the type of fronted adverbial constituent (time adverbials versus others) and the type of subject (pronominal versus nominal subjects).
7 Discussion
American Danish seems to stand out compared to the other Germanic varieties that are spoken as heritage languages in North America with regard to the number of tokens with which the ‘English’ XSV word order appear in main declarative clauses (5–9%, or, 91–154 tokens, depending on whether the outlier speaker ADJ is included). Although this is not a huge number of tokens, no other Germanic heritage variety in North America is reported to display a similar degree of deviation in word order in this specific syntactic context.
The image that we have gained of American Danish word order in non-subject initial main clauses is that, notwithstanding the incidence of non-European Danish word order, it is quite resilient to change, at least based on this specific dataset. With regard to syntactic variation, we see neither a movement towards a fixed word order nor an inclination towards overuse of word order shared by the two languages in contact, e.g. subject-first word order. This is not in line with Polinsky and Kagan’s (2007: 382) predictions that “fixed word order […] becomes a stable feature of heritage language grammars. […] While it would be tempting to motivate the rigid word order found in heritage grammars by universal tendencies in language encoding […] it may simply be that English acts as a source of transfer.”
As to the other possible causes of the word order changes in non-subject initial main clauses, we have observed one syntactic context in which deviance from the homeland syntax is contact-induced, i.e. can definitely be attributed to the influence of English. This is the case when the fronted adverbials contains an English element or is itself an English element: This feature occurs often together with SV word order. However, topics that include or are preceded by an English element are not necessarily followed by SV word order, and SV word order also appears in Danish clauses with no obvious English influence at all. Pioneer work on the triggering effect of a preceding other-language element has been done by Clyne (1967, 1980) and has been taken further by psycholinguistic approaches to language activation in bilinguals (cf. Kroll and Gollan 2014 and references therein). This is a point that is worth further investigation, but for reasons of space we will not pursue this here.
Another striking factor is the effect that adverbials seem to have for the occurrence of SV word order. However, this might not be a matter of the word class itself but instead a secondary effect: Although we lack comparable data, adverbials as initial constituents seem quite frequent in biographic narratives such as those comprising much of our dataset. The significance effect of the adverbials may thus be an effect of frequency and text type.
There are some noteworthy differences between the Danish migrant group and the other Mainland Scandinavian groups: Compared to American Norwegian and American Swedish, where changes in syntax seem to be a quite recent phenomenon (see Larsson and Johannessen, 2015a, Eide and Hjelde, 2015), the divergence from homeland syntax observable in our dataset is documented in the 1960s and 1970s. The earliness of the feature is due to the fact that, in our speaker sample, original immigrant speakers also produce the non-standard syntactic construction XSV. In contrast, Larsson and Johannessen (2015a, b) observe a clear distinction between immigrant speakers and heritage speakers with regard to syntactic stability in that only heritage speakers show divergence from homeland syntax. Our data sample shows no such regularity: The XSV word order in declaratives is produced by immigrant speakers and heritage speakers alike. This suggests that at least this feature cannot be attributed neither to incomplete acquisition nor to attrition, at least not for the Danish immigrants and their descendants. It in turn corroborates the claim that the specific speaker group with its demographic characteristics must be the point of departure for any analysis.
One could highlight the rapid language shift of the Danish group (caused by the low degree of cohesive settlement and low degree of enduring strong networks among the Danish Americans) as a cause for the non-standard production of XSV word order by both immigrant speakers and heritage speakers. However, this would imply that the notions of incomplete acquisition and language attrition can only be applied to tight-knit groups committed to language maintenance, instead of to every migrant group that contains heritage speakers. This would in turn necessitate a specification of the concept of heritage speaker relative to the processes of attrition and incomplete acquisition in order to integrate the amount of exposure to, and output of, the heritage language (see also Sewell, 2015). 10 Considering the differences in the amount of exposure and use of Danish between the adult immigrants in our data sample (note the average emigration age of 18.4) and the US-born heritage speakers, the production of the non-standard syntactic construction by both immigrant and heritage speakers suggests that the relationship between syntactic output and language input is not straightforward. This is in line with, for example, Sewell’s findings on Wisconsin German, where some speakers who report extensive use of and pronounced exposure to both Wisconsin and European German produce the most cases of ADV+SV (Sewell, 2015: 246). Putnam and Sanchez (2013) discuss the role of activation of the heritage language during lifespan as a main factor in the development of heritage grammar. This approach provides a promising perspective that takes into account the language acquisition and language use of heritage speakers. 11
In a contact linguistic perspective, Kühl and Braunmüller (2014) propose a framework of structural stability and divergence in language contact situations. Their framework predicts that stability and divergence respectively, can come about due to, or despite of, language contact. If we apply this to our findings, we see an overall stability of word order in topicalized main declarative clauses and, in particular, the construction with a left-dislocated topic that is resumed by an anaphoric particle – despite the intense language contact situation. We do not see any stabilization of European Danish syntax due to contact (i.e. overuse of certain syntactic constructions as e.g. SV word order in clauses starting with the ADV-CON cognate ‘så’/’so’). Thus, the data shows structural divergence from European Danish due to contact (i.e. the SV word order triggered by an initial English element) yet also structural divergence from European Danish that does not arise from exact replication patterns of the dominant language (i.e. subject-first word order), but is a consequence of the decline of structural norms due to bilingualism and language shift and loss.
8 Conclusion
This paper has presented a study of word order in non-subject initial main clauses in North American Danish. The study is based on the natural speech of 64 speakers (heritage and immigrant speakers) that amounts to a corpus of 103,00 words. Within this corpus, all instances of declaratives with a non-subject fronted element have been taken into account.
Based on analyses of structural and sociolinguistic factors, we have shown that American Danish shows an overall resilience to change in word order, but that nevertheless 5–9% of all instances of non-subject initial declarative clauses show XSV word order (as in English). We have identified linguistic factors that seems to affect the relative order of subject and finite verb, yet, surprisingly, no sociolinguistic factors showed any correlation with the production of XSV and XVS word order, not even the difference between immigrant and heritage speakers (at least not in this specific dataset). Another surprise has been that American Danish seems to be the only Germanic heritage language spoken in North America in which the relative order of subject and finite verb in non-subject initial main clauses displays more than just a few single instances of XSV word order. This makes American Danish different from the other Mainland Scandinavian languages in North America that show but little, and only recent, degree of change, but also from the German varieties that stands out as noticeably stable with regard to word order.
The authors gratefully acknowledge the generous funding given to the project ‘Danish Voices in the Americas’ by the A.P. Møller Foundation, the Carlsberg Foundation, and the Faculties of Humanities at the University of Copenhagen. This study of word order has been conducted as part of this project. We furthermore offer our thanks to two anonymous reviewers and to Janne Bondi Johannessen and Joshua Bousquette for their insightful comments and to our colleagues Jacob Thøgersen and Torben Juel Jensen for statistical advice. All remaining errors remain our responsibility.
References
Aalberse Suzanne and Muysken Pieter . 2013. Language contact in heritage languages in the Netherlands. In Joana Duarte and Ingrid Gogolin (eds.), Linguistic superdiversity in urban areas. Research approaches, 253–274. Amsterdam/Philadelphia: John Benjamins.
Arnbjörnsdóttir Birna . 2006. North American Icelandic: The Life of a Language. Manitoba: The University of Manitoba Press.
Arnbjörnsdóttir Birna . 2015. Reexaming Icelandic as a heritage language in North America. In Janne Bondi Johannessen and Joseph C. Salmons (eds.), Germanic heritage languages in North America: Acquisition, attrition and change, 72–93. Amsterdam/Philadelphia: John Benjamins.
Becker-Christensen Christian . 2010. Nudansk syntaks. Frederiksberg: Samfundslitteratur.
Benmamoun Elabbas , Montrul Silvina and Polinsky Maria . 2013. Heritage languages and their speakers: Opportunities and challenges for linguistics. Theoretical Linguistics 39: 129–181.
Boas Hans C. 2009. The life and death of Texas German. Durham: American Dialect Society.
Bourhis Richard . 2001. Acculturation, language maintenance, and language shift. In Klatter-Folmer Jetske and van Avermaet Piet (eds.), Theories on maintenance and loss of minority languages. Towards a more integrated explanatory framework, 5–38. München a.o.: Waxmann.
Christensen Tanya Karoli . 2010. ’Gud give det bliver til gid’: Nedslag i en sprogforandringsproces. Ny forskning i grammatik 17: 5–22.
Clyne Michael . 1967. Transference and triggering. The Hague: Martinus Nijhoff.
Clyne Michael . 1980. Triggering and Language Processing. Canadian Journal of Psychology 34: 400–406.
Eckert Penelope . 2006. Communities of practice. In Brown Keith and Anderson Anne (eds.), Encyclopedia of language and linguistics. Amsterdam: Elsevier.
Eide Kristin Melum and Hjelde Arnstein . 2015. Verb second and finiteness morphology in Norwegian Heritage Language of the American Midwest. In Page Richard B. and Putnam Michael T. (eds.), Moribund Germanic Heritage Languages in North America: Perspectives and empirical findings, 64–101. Leiden: Brill.
Gehring Charles . 1973. The Dutch language in colonial New York. An investigation of a language in decline and its relationship to social change. PhD dissertation, Indiana University.
Grøngaard Jeppesen Torben . 2005. Danske i USA 1850–2000. En demografisk, social og kulturgeografisk undersøgelse af de danske immigranter og deres efterkommere. Odense: Odense Bys Museer.
Grøngaard Jeppesen Torben . 2011. Scandinavian descendants in the United States. Ethnic groups or core Americans? Odense: Odense Bys Museer.
Hansen Erik and Heltoft Lars . 2011. Grammatik over det danske sprog. Odense: Syddansk Universitetsforlag.
Hasselmo Nils . 1974. Amerikasvenska. En bok om språkutvecklingen i Svensk-Amerika. Lund: Berlingska Boktryckeriet.
Haugen Einar . 1953. The Norwegian Language in America. The American Dialects of Norwegian. Philadelphia: Philadelphia University Press.
Hjelde Arnstein . 2001. A bilingual community and research problems: The Coon Prairie settlement and problems of distinguishing language contact phenomena in the speech of Norwegian Americans. In Sture Ureland (ed.), Global Eurolinguistics. European Languages in North America – Migration, Maintenance and Death, 209–230. Tübingen: Niemeyer.
Hjelde Arnstein . 2015. Changes in a Norwegian dialect in America. In Janne Bondi Johannessen and Joseph C. Salmons (eds.), Germanic Heritage Languages in North America. Acquisition, Attrition and Change, 283–298. Amsterdam/Philadelphia: John Benjamins.
Hopp Holger and Putnam Michael T. . 2015. Syntactic restructuring in heritage grammars. Word order variation in Moundridge Schweitzer German. Linguistic Approaches to Bilingualism 5 (2): 180–214.
Hvidt Kristian . 1971. Flugten til Amerika eller Drivkræfter i masseudvandringen fra Danmark 1868–1914. Århus: Universitetsforlaget.
Johannessen Janne Bondi . 2015. Attrition in an American Norwegian heritage language speaker. In Janne Bondi Johannessen and Joseph C. Salmons (eds.), Germanic Heritage Languages in North America. Acquisition, Attrition and Change, 46–71. Amsterdam/Philadelphia: John Benjamins.
Kroll Judith F. and Gollan Tamar H. . 2014. Speech planning in two languages: What bilinguals tell us about language production. In Goldrick Matthew , Ferreira Victor S. and Miozzo Michele (eds.), The Oxford Handbook of Language Production. Oxford: Oxford University Press. DOI: 10.1093/oxfordhb/9780199735471.013.001
Kühl Karoline . 2015. ’Det er easy at tale engelsk også’. Amerikadansk i 1960’erne og 1970’erne. Nydanske Sprogstudier 47: 39–60.
Kühl Karoline and Braunmüller Kurt . 2014. Stability and divergence in language contact: An extended perspective. In Kurt Braunmüller, Steffen Höder and Karoline Kühl (eds.), Stability and divergence in language contact. Factors and mechanisms, 13–38. Amsterdam/Philadelphia: John Benjamins.
Larsson Ida and Johannessen Janne Bondi . 2015a. Incomplete acquisition and verb placement in heritage Scandinavian. In Page Richard B. and Putnam Michael T. (eds.), Moribund Germanic Heritage Languages in North America: Perspectives and empirical findings, 153–189. Leiden: Brill.
Larsson Ida and Bondi Johannessen Janne . 2015b. Embedded word order in Heritage Scandinavian. In Hilpert Martin et al. (eds.), Current issues in Nordic Languages, 239–264 Berlin/New York: de Gruyter.
Larsson Ida , Tingsell Sofia , Andréasson Maria , Lyngfelt Benjamin , Nilsson Jenny . 2012. Amerikasvenskan för och nu. Norsk Lingvistisk Tidsskrift 30: 263–286.
Louden Mark . 1994. Syntactic change in multilingual speech islands. In Berend Nina and Mattheier Klaus (eds.), Sprachinselforschung. Eine Gedenkschrift für Hugo Jedig, 73–91. Frankfurt am Main: Lang.
Matras Yaron . 2009. Language contact. Cambridge: Cambridge University Press.
Montrul Silvina . 2009. Incomplete acquisition of tense-aspect and mood in Spanish heritage speakers. International Journal of Bilingualism 13 (3): 239–269.
Newman Zelda Kahan . 2015. Discourse markers in narratives of New York Hassidim. More V2 attrition. In Janne Bondi Johannessen and Joseph Salmons (eds.), Germanic heritage languages in North America: Acquisition, attrition and change, 178–197. Amsterdam/Philadelphia: John Benjamins.
Nützel Daniel and Salmons Joseph (2011) Language contact and new dialect formation: Evidence from German in North America. Language and Linguistic Compass 5 (10): 705–717.
Polinsky Maria . 2011. Reanalysis of adult heritage language: A case for attrition. Studies in second language acquisition 33: 305–328.
Polinsky Maria and Kagan Olga . 2007. Heritage Languages: In the ‘Wild’ and in the Classroom. Language and Linguistic Compass 1 (5): 368–395.
Putnam Michael and Sánchez Liliana . 2013. What's so incomplete about incomplete acquisition? A prolegomenon to modelling heritage language grammars. Linguistic Approaches to Bilingualism 3 (4): 478–508.
Quist Pia . 2000. Ny københavnsk ‘multietnolekt’. Om sprogbrug blandt unge i sprogligt og kulturelt heterogene miljøer. Danske Talesprog 1: 143–211.
Romaine Suzanne . 1989 [1995]. Bilingualism. Oxford: Blackwell.
Rothman Jason . 2009. Understanding the nature and outcome of early bilingualism: Romance languages as heritage languages. International Journal of Bilingualism 13 (2): 155–163.
Salmons Joseph (ed.), 1993. The German Language in America 1683–1991. Wisconsin: Max Kade Institute for German-American Studies, University of Wisconsin-Madison.
Sewell Alyson . 2015. Sociolinguistic and syntactic variation in Wisconsin German Narratives. In Page Richard B. and Putnam Michael T. (eds.), Moribund Germanic Heritage Languages in North America: Perspectives and empirical findings, 224–250. Leiden: Brill.
van Marle Jaap . 2001. American ‘Leeg Duits’ (Low Dutch) – a neglected language. In P. Sture Ureland (ed.), Global Eurolinguistics. European Languages in North America - Migration, Maintenance and Death, 79–101. Tübingen: Niemeyer.
There is one exception to this rule. Clauses with clause-initial ‘emotive’ adverbs like gid ‘I wish’ and bare ‘if only’ require the word order subject+verb, for example, gid du ville spise din grød ‘I wish you would eat your porridge’, bare han ville spise sin grød ‘if only he would eat his porridge’ (Becker-Christensen, 2010: 76; Christensen, 2010; Hansen and Heltoft, 2011: 1596). In addition, Copenhagen multiethnolect shows ‘V2 violations’ (as in (1c)) (Quist, 2000: 152–153). To our knowledge, however, the linguistic conditions for divergent word order in this particular variety have not been subject to systematic (corpus-based) analyzes.
No dialectal variation in European Danish explains the variation we see in American Danish.
An anonymous reviewer pointed out to us the paper by Johannessen (2015) on word order in American Norwegian. This is a case study of a single speaker (Daisy) who also is part of the dataset used by Larsson and Johannessen (2015b). As far as we can see, it is Daisy, who produces most of the V2 violations in the main clauses that both papers discuss. For this reason, we do not discuss Johannessen (2015) separately.
Polinsky (2011) defines heritage speakers as follows: “A heritage speaker is a bilingual who grew up hearing and possibly speaking an immigrant language or a minority language as a home language, and who has been dominant in the majority language of the wider community since early childhood.” See also the much-applied and precise definition of a heritage language by Rothman (2009: 156).
These data have been digitalized and transcribed as part of the research project ‘Danish Voices in the Americas’ (2014–2018) at the University of Copenhagen.
For a more comprehensive account on the macro-sociolinguistic background of the Danish migrant group, see Kühl (2015). For a comprehensive comparison of the Mainland Scandinavian immigration to North America, the different settlement and assimilation patterns and their socioeconomic consequences, see Grøngaard Jeppesen (2005) and (2011).
Presentation at the ICLaVE 8 Conference 2015 in Leipzig.
We are as yet unable to measure fluency in this data, but our overall impression is that there is no obvious correlation between fluency, lexical retrieval difficulties, and production of SV word order in declarative main clauses with a fronted non-subject constituent. However, this remains to be investigated.
There may be alternative ways of measuring ‘length’ or ‘weight’, for example by considering word length, e.g. by including counts of number of syllables or phonemes. However, this is not a straightforward task, because it requires decisions to be taken as to whether one should count phonological, ‘expected’, phonemes and syllables or actually realized phonemes and syllables. We have followed Eide and Hjelde’s (2015: 40) procedure of counting words, irrespective of their actual length in terms of phonological structure and/or phonetically reduced pronunciation.
As we do have data from tight-knit Danish American settlements (still in the process of digitalization and transcription), a further study might be able to assess possible differences in the syntactic behavior of individual speakers and speakers who are part of a heritage Danish speaking group.
Another perspective and research interest is represented by Benmamoun, Montrul and Polinsky (2013) whose focus lies on the assessment of the final stages of (incomplete) grammar in heritage speakers.