 Hello. This e-lecture provides an overview of the study of language universals, that is, of linguistic statements that apply to all languages. In particular, we will look at language universals and their position within linguistic research. We will define central types of universals and will then look at the typological parameters that allow us to classify the languages of the world and language universals. Two major methodological approaches have been adopted in linguistics as far as language universals are concerned. They can be contrasted on the basis of two parameters. On the one hand, we can ask the question how many languages should be chosen for universal statements and we have two possibilities. On the one hand, we could look at a large number of languages, that is one plus n languages. On the other hand, we could confine our statements to just one language. Since all languages share the same core, one will suffice would be the argument. A second parameter concerns the formulation of the type of universals, that is the degree of abstractness. What type of universals do we have in mind? We have the choice between concrete universals or relatively abstract universals. That is concrete statements that can be empirically tested and abstract universals relating to general cognition in many cases. The choice of these parameters results in two philosophies concerning universal research in particular and linguistics in general. Namely, empiricism and rationalism by and large associated with two key figures in linguistics, Joseph Greenberg and Norm Chomsky. Let us look at empiricism first. Some linguists have argued that in order to carry out research on language universals, it is necessary to have data from a wide range of languages. This empiricist view is most closely associated with the work of Joseph Greenberg. The resulting universals can be stated in terms of relatively concrete rather than abstract analysis. Let us look at an example. In all these cases, behind me, the subject precedes the object, irrespective of the position of the verb. So we have the sentence, the man sees the woman in all languages represented phonetically. The man sees the woman in German, der Mann sieht die Frau. In Arabic, and in Japanese, So in all these languages we can observe a tendency irrespective of the position of the verb, the subject precedes the object. And if the database is large enough, a general principle emerges, referred to as universal number one by Joseph Greenberg. Well, here it is in declarative sentences with nominal subject and object. The dominant order is always one in which the subject, which we marked red, precedes the object marked in a blue color. Let us contrast this view with the rationalist view. This view, which is most closely associated with the work of Norm Chomsky, claims that the best way to learn about language universals is by the detailed study of one individual language. The resulting universals are abstract rather than concrete and would be explained by the innate character of human language. They are not formulated in terms of concrete statements rather they are formulated as constraints on linguistic operations or as general principles that govern natural language. The following example postulates such a constraint that blocks a contraction operation. In most spoken varieties of English, function words often contract with nearby words. A typical example is the Wanna construction. Now here we have a sentence I want to see marry. If we turn that into an echo question, we get I want to see who. And if we turn the echo question into a WH question by means of fronting the WH element, you get who do I want to see. And eventually you can contract want to to Wanna and the result is something like who do I want to see. Now let's contrast this with our second example. I want marry to see me. And again, we turn that first into an echo question. I want who to see me. And then into a WH question by means of fronting the WH element. Who do I want to see me. And then we contract this want to to Wanna. Who do I want to see me. And the result which most native speakers would support is an ungrammatical sentence where Wanna construction is impossible. The explanation for this phenomenon is quite simple, but it rests on a principle that is part of the theory. It cannot simply be extracted from empirical studies. It rests on the principle that elements that are moved leave behind traces. Traces are syntactic material that is phonologically empty. Whereas the trace in the first case is sentence final and does not interfere with want and to. It blocks Wanna contraction in our second case because it occurs between want and to. So here we do not have an empirical explanation, but one which is based on pure theoretical insights. That is an example of rationalism. Most language typologists, however, favor the empirical approach for the following reasons. The first reason is that languages vary a lot. Wagon language universals has shown a wide range of variation across languages. Evidence based on an individual language can often not justify a universal statement. For example, the implicational universal that subject object verb that is SOV languages have post positions would certainly be held true on the basis of Japanese. But what about Persian and SOV language with prepositions? Well, let's illustrate this first. Here we have the sentence in the two languages, Japanese and Persian. The man sees the woman. So here's the Japanese case first. The man, the woman, see SOV and the same in Persian. Subject object verb. Okay, and what about adpositions? Well, in Japanese we have the book in, in the book. The book in and in Persian in book. So in the book and whereas in Japanese we have a post position. So a clear adherence to the modifier head principle in Persian. This principle modifier head applies to objects and their verbs, but not to prepositions and their nouns. So shall we exclude this language from the universal? Well, statements based on one language are simply too strong because of this variation across languages. A second problem concerns the validation of certain language universals. Often putative language universals crumble as soon as data from other languages arrives. Take the X bar syntax scheme as an example where you have the principle that you have a phrasal category that can be expanded into a specifier and a bar level category and so on. This scheme was proposed as a universal on the basis of English data first where it is indeed the case that for example determiners precede their nouns. But what about other languages? Today this scheme is a general scheme or more or less a tendency rather than a universal and surely it cannot be validated on the basis of one language alone. Finally we have languages which we might want to call exceptional. The one language approach may ignore a number of interesting properties that may as well not be included even within a comprehensive sample of the world's languages. For example, click consonants as regular phonemes are restricted to Koizan and neighboring Bantu languages in Southern Africa. How can they become part of the formulation of universals in a one language approach? Thus detailed work on language universals needs a representative sample of languages that is defined by the absence of the following biases. There should not be any genetic biases that is any language, any language family should be included. There shouldn't be any aerial biases, any region, no restriction on regional aspects and typologically we should take any language type and should not restrict the type of languages on the basis of typological parameters. Let us now look at the various types of universals that can be postulated. Depending on the theoretical background on the one hand and on theoretical considerations about the formulation of language universals on the other, universals can be categorized along several parameters. There are, for example, substantial universals. These are universals in a wider sense. Basically, they are the categories that must be present in each individual human language. For example, the set of open class syntactic categories such as noun, verb, adjective and so on. Formal universals are statements on the form of rules of grammar. In generative grammar, for example, formal universals are expressed in terms of constraints, that is, in terms of mechanisms that delimit the class of possible rules. And then we have non-implicational universals. A large number of properties that are found in the world's languages can be stated without reference to any other property of the given language. These properties are called non-implicational and they can be contrasted with implicational universals, which relate the presence of one property to the presence of some other property. Absolute universals can be classified as exceptionless. So whenever you find something which has no exception, you call it an absolute universal. And this can be contrasted with tendencies, which are also referred to as statistical universals, and which, well, perhaps you agree, seem to run counter to the idea of having universals at all. How can a universal have exceptions? Well, in stating language universals, these types may often be combined. For example, you can have implicational and absolute universals, you can have formal absolute universals, and so on. A more concrete way of classifying the languages of the world uses a set of typological parameters, ranging from phonology to syntax and even beyond. So let's look at them. In principle, one could choose any linguistically relevant parameter along which to typologize languages. However, the range of parameters should be restricted to those along which languages in fact do vary. For example, since all languages have vowels, the presence versus the absence of vowels is no longer of interest for the study of language variation. Let us briefly illustrate some typologically relevant parameters for each linguistic branch. By the way, more details can be found in my e-lecture, the classification of languages. Phonologically, languages can be typologized on the basis of segmental aspects such as their sound inventories, the presence or absence of certain segments, or particular segmental patterns. Other phonological parameters use suprasegmental aspects. For example, the phonological structure of syllables, the rhythmic patterning of words and sentences, or tonal aspects. The most important morphological or structural parameter defines languages on the basis of their correspondences between words and morphemes. There are two extremes. Analytic languages with almost no morphology, that is with a one-to-one correspondence between words and morphemes, and synthetic languages where words may consist of one plus any number of morphemes. The most important parameter along which the languages can be classified syntactically is the order of the functional constitutions of clause structure, that is subject, verb and object. Beyond these major linguistic fields, there are other significant parameters that cross the borderline between linguistics and psychology. Let's look at them. Now, typologizing languages beyond the central linguistic parameters is far more difficult and often less systematic. Furthermore, cultural and environmental aspects may be influential as well. For example, in the Eskimo-Aliut languages, we find more words for snow than in the Indo-European languages. Nevertheless, typologists have discussed several other parameters. One of them is the numerical system that a language uses. So here we have a set of randomly chosen numbers from one to a hundred, and the central question is on the basis of which basic number are complex numbers such as, well, let's take 51 as an example, on the basis of what basic number are such complex numbers built. Most languages, for example English, have a decimal system that is counting and number composition is based on the number 10. So 51 can be morphologically analyzed as 5, 50 and 1, and numerically we have 5 times 10 plus 1. Georgian, by contrast, has a vegesimal system based on multiples of 20. To express a number larger than 20, you perform a calculation based on the 20. So 51 would be orm 2, odds da, 20, ati, 10, erti, so 2 times 20 plus 10 plus 1. Another parameter concerns the use of color words. It has been found that languages can be grouped into a general system ranging from 3 to 7 color terms and more. So these are just two of many other possibilities to classify languages on the basis of non-linguistic parameters. Let us summarize. In this e-lecture we saw that there are two central approaches towards the definition of language universals. The rational approach which generates universals primarily on the basis of the detailed study of one language claiming that the central principles must be common to all languages and the empiricist approach that generates language universals by means of an empirical study of as many as possible languages. Furthermore, we saw that language universals can be characterized in many different ways, such as absolute, implications, and so on. And last but not least, we listed the central parameters that linguists use to classify the languages of the world and eventually to define language universals in the Greenbergian way. If you want to find out more about these issues, I recommend to consult our Language Index where you can apply these parameters to a large corpus of audio-supported language samples and if you wish, where you can contribute with an entry of your own. Thank you.