The studies presented in chapters 5 and 6 use English datasets and most of the discussion and examples provided in this dissertation are focused on English cases. An important question is the extent of the universality of the grammaticalization features discussed so far, given that languages involve major typological differences. For example, the highly grammaticalized end of the cline introduced in section 2.3 involves cliticization and affixation. Yet a traditional distinction between languages is whether they are synthetic (where one word can be made of multiple morphemes) or analytic (where one word tends to consist of one morpheme only).5 Therefore, it is to be expected that advanced grammaticalization processes in analytic languages display different features from those displayed by the synthetic ones, given that cliticization and affixation are rarer (or non-existent) in the former.
In a similar fashion, subject-verb-object word order is also a main way to categorize languages. Section 2.2.1 introduced the parameter of syntagmatic variability (i.e. how an element can be shifted around its context) where more grammaticalized items tend to have less syntagmatic variability. Since word order can also be relatively fixed or free from one language to another, this may also translate differently when it comes to grammaticalization processes and syntagmatic variability.
The view adopted in the following chapters is that the parameters of grammaticalization discussed in sections 2.2.1 and 2.2.2 are relevant to most major languages. The task of comparing grammaticalization across all languages is complicated for obvious reasons, but a
5 Note that the distinction between synthetic and analytic is gradual, since languages can have both synthetic and
analytic ways of expressing certain concepts. For example, comparison in English can be expressed in an analytic manner (e.g. more intelligent) or in a synthetic one (e.g. smarter). This is why the distinction between synthetic and analytic involves fuzzy boundaries.
notable work is Heine and Kuteva (2002) which contains a list of grammaticalization processes from a broad variety of languages. As noted in Narrog and Heine (2011: 13), studies of grammaticalization have mostly been focused on major languages of the world, in particular English and European languages.
There are examples of grammaticalization that are fairly similar among European languages, such as future auxiliaries developing out of motion verbs. These illustrate a move from concrete meanings (e.g. physical motion) to more abstract ones (e.g. future tense), which is a known characteristic of grammaticalization (sections 2.2.1 and 3.3). A straightforward example in English is the motion verb go, which developed into the be going to + infinitive future construction, which can be used to express future actions that do not necessarily involve physical movement (e.g. I’m going to sing). The same use exists in French and Portuguese with the verbs aller and ir (to go) which can form sentences such as je vais chanter and Vou cantar (I’m going to sing) which do not imply physical motion. In fact, most Romance languages also have this use of the motion verb go. Despite such similarities, there are also notable exceptions that question to what extent grammaticalization “behaves” in the same way in different languages, in particular when considering different language families.
In section 2.1, phonological attrition was presented as a rather common feature in secondary grammaticalization processes, with mention to Norde (2012) and the example of the development of the possessive pronoun (h)inn into the suffix -en in Swedish. There are also straightforward examples in English such as the development of will into ’ll, or that of going to into gonna. However, Chinese is an example of a language where grammaticalization involves little form change (Bisang 2009), which means that phonological reduction is not a prominent parameter at play in this instance.6 Furthermore, many other East and mainland Southeast Asian languages also have highly grammaticalized items that do not involve phonological attrition (Bisang 2011: 110-113). Similarly, in a broad study involving 19 languages from genetically unrelated families, Schiering (2006) shows that in the case of clitization, phonological attrition is not a defining property of grammaticalization in all the languages involved in that study. Another salient feature of grammaticalization known as semantic bleaching (sections 2.2.1 and 3.3) is also problematic in Chinese as it involves many cases of grammaticalization that show an accretion of more meaning instead (e.g. Xing 2015).
This shows that while some aspects of grammaticalization might seem universal, there can be exceptions in other languages. The parameters used in the subsequent English studies
6 Note that reductions in syllable duration and vowel quality/duration are attested, as reported in Ansaldo and
are presented in chapter 3 and there is no doubt that their prominence changes from one language to another. However, the models introduced in chapter 4 will show that the relevance of these parameters can be calculated on the basis of empirical data, a concrete application of which is presented in chapters 5 and 6. This is one of the advantages of these models because it means that when applied to a language different from English, the parameters themselves can be recalculated for that language specifically. This also enables comparison between different languages as the prominence of specific parameters of the models can be compared. This possibility will not be explored in the present dissertation, but might be a fruitful endeavour for future works.
Attention has also been given to processes of grammaticalization in situations of language contact. For example, the development of creoles and pidgins is of interest since creoles in particular are known to have grammars in which complexity is greatly reduced (McWorther 2001, Hopper and Traugott 2003: 212-230). Another potential source of grammaticalization processes is bilingualism in situations of language contact (Matras 2011). While these specific situations are not dealt with in the present dissertation, they illustrate that there are situations in which grammaticalization processes must inherently be studied by taking different languages into account.
When discussing grammaticalization across languages, some researchers have undertaken the task to compare similar processes across a set of languages, often from the same family, and to consider their respective degrees of grammaticalization. For instance, Lamiroy and de Mulder (2011) discuss auxiliaries and other products of grammaticalization across three Romance languages, namely French, Italian and Spanish. They state that French is more grammaticalized than Italian, which in turn is more grammaticalized than Spanish. They therefore rank entire languages across a grammaticalization cline. This approach is rather uncommon, as noted by Narrog and Heine (2011: 14). However, this shows that there are different interpretations of what constitutes a grammaticalization cline and what the elements on that cline can be. The next section deals with a similar concern as it discusses which elements are undergoing grammaticalization by presenting the construction-based view according to which these elements are constructions, as opposed to single lexical or grammatical items.