Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, LINGUISTICS (linguistics.oxfordre.com). (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 16 August 2017

Compounding in Morphology

Summary and Keywords

Compounding is a word formation process based on the combination of lexical elements (words or stems). In the theoretical literature, compounding is discussed controversially, and the disagreement also concerns basic issues. In the study of compounding, the questions guiding research can be grouped into four main areas, labeled here as delimitation, classification, formation, and interpretation. Depending on the perspective taken in the research, some of these may be highlighted or backgrounded.

In the delimitation of compounding, one question is how important it is to be able to determine for each expression unambiguously whether it is a compound or not. Compounding borders on syntax and on affixation. In some theoretical frameworks, it is not a problem to have more typical and less typical instances, without a precise boundary between them. However, if, for instance, word formation and syntax are strictly separated and compounding is in word formation, it is crucial to draw this borderline precisely. Another question is which types of criteria should be used to distinguish compounding from other phenomena. Criteria based on form, on syntactic properties, and on meaning have been used. In all cases, it is also controversial whether such criteria should be applied crosslinguistically.

In the classification of compounds, the question of how important the distinction between the classes is for the theory in which they are used poses itself in much the same way as the corresponding question for the delimitation. A common classification uses headedness as a basis. Other criteria are based on the forms of the elements that are combined (e.g., stem vs. word) or on the semantic relationship between the components. Again, whether these criteria can and should be applied crosslinguistically is controversial.

The issue of the formation rules for compounds is particularly prominent in frameworks that emphasize form-based properties of compounding. Rewrite rules for compounding have been proposed, generalizations over the selection of the input form (stem or word) and of linking elements, and rules for stress assignment. Compounds are generally thought of as consisting of two components, although these components may consist of more than one element themselves. For some types of compounds with three or more components, for example copulative compounds, a nonbinary structure has been proposed.

The question of interpretation can be approached from two opposite perspectives. In a semasiological perspective, the meaning of a compound emerges from the interpretation of a given form. In an onomasiological perspective, the meaning precedes the formation in the sense that a form is selected to name a particular concept. The central question in the interpretation of compounds is how to determine the relationship between the two components. The range of possible interpretations can be constrained by the rules of compounding, by the semantics of the components, and by the context of use. A much-debated question concerns the relative importance of these factors.

Keywords: definition, head, synthetic compound, exocentric compound, copulative compound, Recoverably Deletable Predicate, skeleton and body, Parallel Architecture, onomasiological approach

Compounding is a word formation process based on the combination of lexical elements. The elements can be characterized as words, stems, or lexemes, depending on the language and on the theoretical framework adopted. In the theoretical literature, the discussion of compounding is marked by disagreement on basic issues. Here, these issues will be grouped into four main areas, labeled as delimitation, classification, formation, and interpretation. To each of these, a section will be devoted. Not all of the issues are equally important in all theoretical frameworks and perspectives. A question that concerns all of them is to what extent theories can be applied crosslinguistically. The final section gives some general considerations about the relationship between the different issues. As much of the discussion in the literature focuses on English, English compounds will be central in the presentation of issues. However, examples from other languages will be included as well, in particular when they raise issues that do not arise in English.

1. Delimitation

A wide range of criteria have been used to characterize compounds and distinguish them from other phenomena, in particular syntactic phrases and derivations. The preference for certain criteria is determined by three questions, listed in (1).

(1)

Compounding in Morphology

A typical compound in English is textbook. It has a range of properties that can be evaluated for the three factors in (1). Morphologically, it consists of two uninflected nouns. Phonologically, it has a characteristic stress pattern. Orthographically, the two components are written together. Syntactically, it behaves as a noun. Semantically, it refers to a type of book, marked in some way by text.

When these properties are considered in the light of (1), a good example of (1a) is the question of whether orthographic criteria are acceptable. In many linguistic theories, starting from Saussure (1916), it is assumed that orthography is not itself part of language. Such considerations have to be distinguished from the question how useful the criterion is. It is well known that in English, the orthography of many compounds is variable. Lieber and Štekauer (2009, p. 7) give the example of flower pot, flower-pot, flowerpot. This makes it difficult to decide whether it is a compound on the basis of this orthographic criterion, a problem of type (1c). Whereas in English, compounds are most often written as two words, in other Germanic languages compounds are generally written together, which highlights the issue in (1b). The combination of problems involved has led to a fairly general rejection of orthography as a criterion for compoundhood.

Phonological criteria offer a good example of the problems involved in a positive answer to (1b). The Dutch examples in (2) illustrate contexts for final obstruent devoicing.

(2)

Compounding in Morphology

The final consonant of the stem in (2a) is realized as /t/ at the end of a word, but when it is followed by an affix starting with a vowel, whether inflectional as in (2b) or derivational as in (2c), resyllabification takes place and the final consonant is realized as /d/. Dutch orthography represents the underlying /d/ in all cases. Compounds as in (2d) do not have resyllabification, so that the final obstruent is devoiced and realized as /t/.

A criterion for recognizing compounds based on final obstruent devoicing would necessarily be language-specific. In English, the rule does not apply. In other languages where it applies, there is no reason to assume that it works in exactly the same way as in Dutch. Moreover, in many Dutch compounds, the rule does not have any observable effect. In hoofdkussen (lit. ‘head cushion’; i.e., pillow), the context for resyllabification does not exist. In kunstacademie (‘art academy’), the first component does not have an underlying voiced final obstruent. This reduces the value of the criterion in case of a positive answer to (1c). Similar considerations apply to criteria based on stress (cf. Giegerich, 2009).

Morphosyntactic criteria take as their starting point the structure of the compound. Usually, a structure such as (3) is assumed.

(3)

Compounding in Morphology

Adopting the binary structure in (3) means that compounds with more than two basic components, for example church history textbook, are the result of recursion, in this example [[church history] [textbook]]. For certain types of compound, for example philosopher-singer-songwriter, such an analysis is rather unnatural, because the choice between the possible orders of combination is arbitrary. However, (3) gives a good basis for discussing morphosyntactic properties of (most) compounds. In English and other Germanic languages, Y is usually the head of the compound.

One relevant issue is the status of X and Y. Especially in the German tradition, there is a tendency to identify intermediate categories between stem and affix. Fleischer (1969, pp. 63–66) is an influential early source on this, but he refers to older sources. Often affixoid is used for elements such as ‑man in postman or ‑ful in careful. This approach is not compatible with a positive reply to (1c), but rather introduces a cline from more stem-like to more affix-like elements, resulting in more compound-like or more derivation-like words.

Another issue pertaining to X and Y in (3) is to what extent they can be inflected. The problem can be illustrated with the plural inflection in Dutch (4) and Italian (5).

(4)

Compounding in Morphology

(5)

Compounding in Morphology

In (4), there are two singular nouns (Z) formed with the same components, but X is singular in (4a) and plural in (4b). They have a different meaning in the sense that (4a) designates a council for a city, whereas in (4b) the council has its scope over several cities. Both nouns are generally considered compounds in Dutch. In (5), the contribution of X and Y is reversed. In Italian, N+N compounds are left-headed. As a consequence, the plural ending can either attach to the head X, as in (5b), or to the full compound Z, as in (5c). Dictionaries tend to give (5b), but grammars give both forms, for example, Dardano and Trifone (1985, p. 120). If (4) and (5) are compounds, meaningful inflection of components as well as inflection of the compound attaching to the first component must be accepted.

A slightly different case is genitive inflection. Some relevant examples from different languages are given in (6).

(6)

Compounding in Morphology

German (6a) is generally considered a compound. It should be noted, however, that Friedens is the genitive form of Frieden (‘peace’). Traditionally, the ending ‑s is analyzed as a linking element in German grammar. Historically, many linking elements have their origin in genitive endings. Synchronically, however, there are some frequent combinations that do not correspond to genitives, as in (7).

(7)

Compounding in Morphology

As illustrated in (7), all nouns in ‑ung and ‑heit have a linking element ‑s when they appear as X in a structure such as (3). The genitive form of these nouns is the same as the nominative, without ‑s, and they do not have ‑s as an inflectional ending in other forms either. On the status of linking elements and their relation to syntax, cf. Koliopoulou (2014).

The English genitive in (6b) is ambiguous. As noted by ten Hacken (1994, pp. 81–83), the ambiguity can be resolved in various ways. One of them is by means of agreement with the determiner, as in (8).

(8)

Compounding in Morphology

As indicated by the brackets, this in (8a) determines the bracketed complex expression α‎, whereas these in (8b) is part of the noun phrase β‎. Therefore, (6b) may be considered a compound in contexts such as (8a) but not in contexts such as (8b). From Marchand’s (1969, pp. 65–69) discussion, it is obvious that the genitive construction illustrated in (8a) is particularly common with human or at least animate nouns as X.

If Polish (6c) is considered as a compound, it is left-headed. However, traditional Polish grammar considers it as a lexicalized syntactic construction (e.g., Szymanek, 2010, p. 218). Observing the parallelism with the English construction, ten Hacken (2013a) argues for an analysis as a compound, while noting the same ambiguity as in (8).

A third interesting case is that of relational adjectives (RAs). The way inflection plays a role in RA+N constructions is illustrated for Polish in (9).

(9)

Compounding in Morphology

The RA autobusowy agrees with the noun it modifies in case, number, and gender, as illustrated by the contrast in (9). When the nominative in (9a) is changed to the genitive in (9b), the ending ‑y on the RA is replaced by ‑ego. For Matthews (1974, p. 35), who gives Latin examples, this is a sufficient reason to exclude RA+N constructions from compounding. To the extent they are fixed expressions, he assigns them to lexicography, analyzing them as idioms. For Polish, Szymanek (2010, pp. 218–219) adopts a similar position. The opposite view, defended, for instance, by Levi (1978), appeals to semantic arguments.

As an example of a criterion based on the relation between the components X and Y and the compound Z in (3), Di Sciullo and Williams (1987) adopt the Right-hand Head Rule (RHR) not as a generalization over compounds but as a delimiting criterion for word formation, which can be used to distinguish compounding from syntax. According to the RHR, Y in (3) is the head of the compound by virtue of its being the right-hand component. Z inherits the syntactic category and other properties from Y. This is illustrated in German (10).

(10)

Compounding in Morphology

In (10), gender is indicated for each of X, Y, and Z. As the two components have different genders, it can be seen that Z takes its gender from Y, not from X. Especially in Germanic languages, taking the RHR as a criterion yields results that coincide with general usage of the term compound. However, (11) gives some examples of expressions that are excluded by this criterion.

(11)

Compounding in Morphology

The French expression in (11a) can be interpreted as evidence that compounds in French are left-headed. The same is true for Italian in (5). The entire compound shares its masculine gender with centre, whereas ville is feminine. However, Di Sciullo and Williams (1987, p. 83) interpret such observations as evidence that French does not have compounds. The Dutch example in (11b) is an exocentric compound. It fails the RHR in the sense that its gender is neuter, whereas kant (‘side’) is nonneuter.

A purely syntactic criterion for the distinction between compounds and phrases can be based on the observation, originally by Postal (1969), that the nonhead of a compound, X in (3) in languages like English, cannot be the antecedent of a pronoun. This is illustrated in the Dutch example (12).

(12)

Compounding in Morphology

The only masculine noun that hij can refer to in (12) is berg (‘mountain’), but because it is embedded in the compound bergdorp (‘mountain village’), it is not available as an antecedent. An explanation for this impossibility is that nonheads of compounds are interpreted in a generic sense. However, proper nouns, which cannot be generic, also appear as nonheads. They are available for pronominal reference. A Dutch example is (13).

(13)

Compounding in Morphology

Ward et al. (1991) claim that pronominal reference to the nonhead of a compound is governed by pragmatic rather than morphosyntactic constraints. They give a list of real-life examples from English. The majority of them are with proper names; others seem to be performance errors or conscious play on the rule. A typical example of the latter is the dialogue in (14).

(14)

Compounding in Morphology

The contrast between B and B’ in reply to A in (14) is not equally strong for all speakers of English, but for most speakers it is easily detectable. In order to explain that the nonhead of a compound is either generic or a proper noun, ten Hacken (1994, p. 73) proposes that it is not possible to use outside context for the interpretation. Proper nouns like Brahms in (13) refer to an individual directly, but common nouns like berg in (12) need a determiner to identify an individual object. This idea is corroborated by the observation that first names are only possible if they are by themselves sufficient to identify an individual, as illustrated in (15).

(15)

Compounding in Morphology

Compared to the properties discussed so far, semantic aspects have not been used a great deal in the delimitation of compounding. Their role is rather in formation and interpretation. Ten Hacken (2013a), however, proposes to take the way an expression gets its meaning as the main criterion for identifying compounds. The relationship between the components of a compound is not determined by the word formation rule, but emerges from the meaning of the individual components and the use of the compound for naming a concept (cf. section 4). This results in the inclusion of genitive constructions such as (6b–c) as well as RA+N constructions as in (9) in compounding. An interesting case is prepositional constructions such as (16) in Romance languages.

(16)

Compounding in Morphology

The Italian linguistic tradition is divided over constructions such as (16b). Scalise (1992) does not mention them in his overview of compounds, but in their overview of theories, Konecny and Autelli (2015) mention several accounts in which such constructions are given a status close to compounding. The preposition has a function similar to the genitive, and the P+N could be called a periphrastic genitive.

After this overview of criteria that have been used, it is worth reconsidering the questions in (1). The centrality of theory in the selection and formulation of criteria is an obvious given. Dressler’s (2006) objections to existing definitions that they are theory-specific and not crosslinguistically valid are not compatible with a terminological understanding of the nature of definitions. Definitions cannot be refuted. The question is not whether a definition is accurate but whether it delimits a useful theoretical concept. In a pretheoretical sense, it is no problem to use a language-specific definition with criteria that are easy to apply. If the concept of compounding is not used as a basis for theoretical claims, a series of typical properties for their identification, leading to a prototype with gradually decreasing typicality, is sufficient.

2. Classification

In classification, a fundamental distinction has to be made between taxonomies and incidental classes. In a taxonomy, a class of items, for instance compounds, is partitioned into subclasses in such a way that every item of the class belongs to exactly one subclass. There is no such requirement for incidental classes. An example of an incidental class is Roeper and Siegel’s (1978) class of verbal compounds. They define verbal compounds as compounds in which the head is deverbal and has one of the suffixes ‑er, ‑ing, or ‑ed. There is no obvious way the remaining set of compounds would be characterized, except as the complement of this class. In general, taxonomies are theoretically more valued. However, the value depends on the use made of the classes. A division of English compounds into solid compounds, hyphenated compounds, and spaced compounds on the basis of their orthography is of little use, because not much follows from membership in one of the classes.

A widely used classification scheme for compounds is the one found in Sanskrit grammars such as Whitney (1879, pp. 424–456). The three main classes he distinguishes are copulative compounds, determinative compounds, and secondary adjective compounds. At this level, the Sanskrit classification system is a taxonomy. Its principal use is to describe the types of compound found in Sanskrit. This means that it is language-specific. At lower levels, there are also incidental classes, that is, special classes whose complement is not a described class or set of classes. Bloomfield (1933, p. 235) mentions several of these classes, but for the three that remain as the central ones he gives the Sanskrit names dvandva, tatpurusha, and bahuvrihi. The equivalents he gives for them are copulative, determinative, and exocentric. Determinative compounds are exemplified by most of the examples in section 1, for example, (4), (5), and (7).

The idea that bahuvrihi is equivalent to exocentric is somewhat problematic (cf. Koliopoulou, 2015). In Sanskrit, the word bahuvrīhi (lit. ‘much rice’; i.e., rich) exemplifies a class of adjectives that may become nouns by conversion. Two examples from Dutch are given in (17).

(17)

Compounding in Morphology

As opposed to Sanskrit examples, the Dutch examples in (17) are only nouns. (17a) has an A+N, (17b) an N+N structure. While (17) exhibits exocentric compounds of the bahuvrihi type, the classification into endocentric and exocentric is based on different criteria than the ones for the main Sanskrit classes. Bloomfield exemplifies exocentric compounds with turnkey, which is clearly not a bahuvrihi compound.

The distinction between endocentric and exocentric compounds was introduced by Bloomfield (1933, p. 235) and is one of the most widely used taxonomic distinctions among compounds. Nowadays usually understood as based on headedness, for Bloomfield it is a matter of whether the compound belongs to the same (form) class as the head. Thus he calls turnkey exocentric because “the head member is an infinitive verb, but the compound is a noun” (1933, p. 235). The difference between these two interpretations is particularly notable in copulative compounds, as illustrated in (18).

(18)

Compounding in Morphology

In English, copulative compounds of the type illustrated by α‎ in (18a) only occur as the nonhead of compounds. Dutch also has them for adjectives in contexts such as β‎ in (18b). Haeseryn et al. (1997, p. 731) treat β‎ as a compound, not the entire expression of (18b). In Modern Greek, N+N compounding of this type, illustrated in (18c), is productive. In these cases, Bloomfield would call α‎, β‎, and (18c) endocentric. However, as it is impossible to distinguish between a head and a nonhead in these cases, such copulative compounds are not usually considered endocentric in modern interpretations of the term.

Whereas from a language-specific Sanskrit perspective, the classification of dvandva, tatpurusha, and bahuvrihi may be a taxonomic one, the concepts need to be extended to maintain this status for a wider range of languages. As mentioned, the extension of bahuvrihi to V+N constructions such as turnkey was adopted by Bloomfield (1933, p. 235). This construction is very common in Romance languages, as illustrated in (19).

(19)

Compounding in Morphology

It is not generally accepted that (19) should be classified as compounds. Di Sciullo and Williams (1987) and ten Hacken (2010) adopt alternative analyses.

Perhaps the clearest illustration of the difficulty of extending the Sanskrit taxonomy to a wider range of languages is Bauer’s (2008) crosslinguistic study of dvandva. Whereas the examples in (18) all correspond to the Sanskrit class of dvandva compounds, Bauer ends up with a much more detailed typology, including related types of copulative compounding not found in Sanskrit. In this typology, five types of dvandva are distinguished, as well as four types of coordinated compounds that are not dvandva.

A type of compound that has been discussed controversially is synthetic compounds. The term was introduced by Bloomfield (1933, p. 231). Although the precise boundary is not set in stone, usually two types of expression are classified as such, illustrated in (20).

(20)

Compounding in Morphology

All examples in (20) show a combination of two stems and an affix. In (20a–b), the second stem is verbal. Bloomfield (1933, p. 232) calls these semi-synthetic. Marchand (1969, p. 31) uses the term verbal-nexus compound. In these cases, the noun with a V+affix structure can be considered the head of a compound. The nonhead may be the object of the verb, as in (20a), or have a different function, as in (20b). In (20c–d), the second stem is nominal. Semantically, it combines with the adjectival nonhead before combining with the affix. In (20d), poster is not a constituent of the word. On this basis, ten Hacken (2010) argues that they are not compounds but derivations based on a phrase. He also points out the similarity in meaning structure of synthetic compounds of the type in (20a–b) and exocentric compounds as illustrated in (19).

Scalise and Bisetto (2009) give a detailed overview of compound classifications and propose a taxonomy with three layers. At the highest level, they distinguish subordinative, attributive/appositive, and coordinative compounds. Subordinative compounds are divided into ground compounds and verbal-nexus compounds. Attributive compounds and appositive compounds are the divisions of the second class. The five classes resulting from the distinctions on the first two levels of the taxonomy are illustrated in Table 1.

Table 1: Classification of compounds by Scalise and Bisetto (2009)

SUB

ground

windmill

verbal-nexus

bookseller

ATAP

attributive

high school

appositive

swordfish

COORD

coordinative

poet-doctor

A problem with the division in Table 1 is that it is supported only by rather vague descriptions of the labels. The number of examples they give is very limited. There are more examples from an earlier version of the classification, but it is not clear to what extent these are meant to fall into a different class in the new version. Thus, Scalise and Bisetto (2009, p. 46) list girlfriend as attributive but woman doctor as coordinative.

For the third level of the taxonomy, Scalise and Bisetto (2009) refer to the distinction between endocentric and exocentric. Each of the five classes in Table 1 is further subdivided according to this opposition. Technically, this means that the headedness criterion is orthogonal to the classification in Table 1. Although Scalise and Bisetto (2009, p. 50) present the classification as a tree structure, a matrix would be a better representation. The examples in Table 1 are all endocentric. In order to implement this matrix structure, a further column with exocentric examples would have to be added.

A general question about classification is for what purpose a classification is needed. Sanskrit grammarians used their classification as a way of systematically presenting the range of compound constructions in their language. Modern examples of this type of use are found in Marchand (1969) for English and de Haas and Trommelen (1993) for Dutch. They tend to select criteria from the ones presented in this section and supplement them with criteria such as syntactic category and semantic relation between the head and nonhead. There is no reason to expect that the details of such a classification, which at least for the lower levels is arrived at by a bottom-up approach, would be applicable crosslinguistically. Unless theoretical claims are associated with individual classes of compounds, the details of the classification are of limited theoretical relevance.

3. Formation

Whereas the tradition of the delimitation and classification of compounding reaches back to the Sanskrit grammarians, the question of the formation of compounds only came up with the emergence of generative grammar. A landmark in this respect was the 1960 PhD dissertation by Robert Lees. Lees (1960) proposes a rule system that derives a compound from a deep structure that may also serve (with at most minor modifications) as the basis for a sentence. He gives the examples in (21) (1960, p. 117).

(21)

Compounding in Morphology

Lees (1960) did not intend that (21a) is derived from (21b), but that (21a) and (21b) are derived from the same deep structure by means of a different sequence of transformations. As described in more detail by ten Hacken (2009), Lees’s (1960) account triggered extensive discussion. On the one hand, the derivation of compounds from a deep structure that specifies the relation between the two components has the potential to account for the meaning of the compound. On the other, the transformations required to derive a compound from such a richly specified deep structure are so powerful that they undermine the explanatory value of the theory. The emphasis on the advantages or disadvantages of this type of approach was correlated with the position taken in the debate on the place of semantics in generative grammar in the late 1960s and the 1970s.

Researchers embracing generative semantics tried to save the central insight of Lees (1960) that the formation of compounds could account for their meaning. In this vein, Levi (1978) proposed a set of Recoverably Deletable Predicates (RDPs). Instead of the unlimited freedom of describing the meaning of compounds exhibited in (21), RDPs offer a finite set of possible relations between the two components of a compound. The deep structures underlying (21a) and (21c) in Levi’s account would be something like (22).

(22)

Compounding in Morphology

Whereas in the system illustrated in (21), the number of possible deep structures is not limited, the system illustrated in (22) has nine RDPs, three of which can be used in two directions (X make Y or Y make X), so that windmill and flour mill are exactly 12-way ambiguous.

The opposition to Levi’s (1978) system was mainly directed at the generative semantic framework she adopted. However, there were also objections based on the problems of connecting the formation of compounds with their interpretation. Thus, the high degree of ambiguity that is predicted by Levi’s account does not correspond to the perceived meaning. Windmill and flour mill do not have twelve meanings from which one is selected.

As an alternative, Allen (1978) proposed the Variable R Condition. Her hypothesis is that the relation between the two components of a compound depends only on the semantics of the components, not on any constraints imposed by the formation rules for compounding. The different relations in (21) are determined by the slots opened by mill and the ways wind and flour can fill these slots. As a consequence of this hypothesis, the formation of compounds can be studied independently of their interpretation. An example of the resulting approach is Selkirk’s (1982, p. 16) set of rewrite rules for the formation of compounds in English, given in (23).

(23)

Compounding in Morphology

The categories in {} in (23a–b) are alternatives. Reflected in (23) is the Right-hand Head Rule (RHR), which was already encountered in a different role in section 1; cf. (10). For Allen and Selkirk, the RHR is the only generalization contributed by the rule of compounding. The selection of categories in (23) is in principle language-specific. Following Marchand (1969, pp. 100–107), Selkirk (1982, pp. 16–17) analyzes verbs such as mass-produce and dry-clean as back-formations, based on compounds such as mass production and dry cleaning. Therefore, they do not require an extension of the rule in (23c).

Selkirk’s proposal has been influential in particular in inspiring the framework of Distributed Morphology (DM). DM was first proposed by Halle and Marantz (1993). It does not include a lexical component in which word formation could take place, but distributes the workings of morphology between feature bundles, active in syntax; a vocabulary list, used between syntax and PF; and an encyclopedia, consulted between LF and semantic interpretation. Harley (2009, p. 129) describes the goal of DM as developing “a fully explicit, completely syntactic theory of word formation.”

When rules of the type in (23) are accepted as the basis of compounding, the formation and the interpretation of compounds are separated. Perhaps the only type where the formation can still be correlated with the interpretation is verbal compounds. Roeper and Siegel (1978, p. 199) restrict this class to compounds involving the affixes illustrated in (24).

(24)

Compounding in Morphology

For this class, Roeper and Siegel (1978) propose a special mechanism that exploits the argument structure of the verb embedded in the nonhead. Although they formulate this mechanism as a transformation, in later elaborations (e.g., Selkirk, 1982), it was conceptualized as an operation on the argument structure in the lexicon. After the end of research in generative semantics, research on compounding was largely focused on verbal compounds for some decades. Thus, in his chapter on compounding, Spencer (1991, pp. 309–349) devotes 20 pages to verbal compounds as against 5 to root compounds.

The explicit restriction to the three affixes in (24) has been criticized. Thus, Allen (1978) argues that population growth should be treated in the same way. However, Marchand (1969, p. 349) gives growth and spilth as the only deverbal nouns in ‑th. When such unproductive suffixes are considered relevant as well, this raises the question where to draw the line. If bicycle theft and bicycle thief are treated in the same way, the result may be a mechanism in which theft and thief are analyzed on the basis of their meaning, independently of the structure of the form, an approach very similar to the one adopted in generative semantics. A generalization of this approach can be observed in Booij’s (2010) construction morphology. He proposes to dissolve compounding as a category into a hierarchy of more and less inclusive constructions, where the more specialized constructions inherit properties from the more inclusive ones they are specializations of.

Finally, the exact nature of the elements combined in a compound must be considered. Whereas in English, rewrite rules such as (23) take a form as their basis that is at the same time identical to the stem and to the word, other languages show variations in the form. Some Dutch examples illustrating this are given in (25).

(25)

Compounding in Morphology

In (25), the linking elements are indicated in small caps in the literal gloss. In (25a), there is no linking element. (25b–c) illustrate two forms of a linking element that, for standard Dutch and most dialects, only differ orthographically, because the <n> in (25c) is not pronounced. The other widespread form of the linking element is illustrated in (25d). (25e) shows a third linking element, which only occurs with a very limited number of stems. The choice of the form of the linking element depends on the first component, although there exist cases in which the same noun takes different linking elements in different compounds, for example, kindsdeel (lit. ‘child-s-part’; i.e., statutory portion of an inheritance going to a child of the deceased).

A different type of form variation is found in Modern Greek. Here, components of compounds can be stems or words. The contrast is illustrated in (26), taken from Ralli (2013).

(26)

Compounding in Morphology

In (26), the compounds are followed by the base form of their components. In both examples, the thematic vowel of the first component is replaced by the linking element ‑o‑. This means that these components are stems. In (26a), also the second component σπιτ‎- (‘house’) is used as a stem. However, in (26b), it is the entire word ταβέρνα‎ (‘tavern’) that becomes the second component of the compound.

In several contexts, a striking disparity can be observed between heads and nonheads of compounds. Linking elements generally depend on the nonhead. Whereas the status of Greek heads can vary, as illustrated in (26), the nonhead is always a stem. (18) showed that dvandva compounds in English are possible as nonheads of a compound. In other contexts they do not occur. All of these cases suggest that nonheads of compounds permit the appearance of elements that are not possible elsewhere. It is not surprising, then, that there are also phrasal compounds, namely compounds with a phrasal nonhead, for example a connect-the-dots puzzle. Wiese (1996) also mentions extralinguistic material that can appear in nonheads of compounds, for example the # key. Ten Hacken (2003) proposes that this is possible because the nonhead of a compound does not have a syntactic category. For a recent discussion of phrasal compounds, comparing English and German data, cf. Trips (2016).

An even more radical difference in the nature of the elements can be found in the case of neoclassical compounds, for example, anthropomorphism. As Bauer (1998) shows, these formations raise different issues compared to the more typical compounds discussed here. These issues include the status of anthropo and morpho (or anthrop and morph) as elements of the English lexicon and of the rule that combines them. Ten Hacken (2012) proposes an explanation of their emergence in English and a system of formation rules in which they are minimally different from regular compounds. Panocová (2015) compares the status of neoclassical compounds in English and in Russian, concluding that English has a formation system for them, but in Russian they are borrowings.

The formation mechanism for compounds is for most languages relatively simple. Complications arise in particular when the formation of compounds is linked directly with their interpretation. Form-related questions going beyond the combination patterns of components, for example, the presence and shape of linking elements, require language-specific solutions.

4. Interpretation

In early overviews of compounding, it was not so much the process of compounding but the interpretation of the compounds which was the central concern. The crucial question in interpreting a compound is how the relation between the two components should be established. Jespersen (1942) gives an extensive catalogue of relations, while admitting that no such list can pretend to be exhaustive. Lees’s (1960) transformational account instrumentalized these relations as the basis for the deep structure of the compounds, cf. (21). This idea was taken up by Marchand (1969), although the two disagreed on the precise significance of the deep structures (cf. ten Hacken, 2009, pp. 60–63 for an overview of their discussion).

Levi’s (1978) RDPs, illustrated in (22), can be interpreted as a compromise between the constraints of a mechanism for the formation of compounds and the requirements of a mechanism for their interpretation. As a formation mechanism, it has the disadvantage of making each compound 12-way ambiguous. As an interpretation mechanism, it has the limitation that with only 12 RDPs the description of the relation is very crude. A well-known example is the contrast in (27).

(27)

Compounding in Morphology

Both compounds in (27) have for as their RDP, although the intended effect on the referent of the nonhead is opposite. As noted by Downing (1977, pp. 825–828), it is possible to reduce the relations between components of a compound to a finite set, but only as a classification, not as an adequate description of the relation. The main problem of Levi’s approach is perhaps that she tries to derive the relation between the two components exclusively from rules for compounding.

Allen’s (1978) Variable R Condition takes the opposite position. The Variable R Condition says that the relation R between the two components of a compound is only governed by the semantics of the components. The head component opens a range of slots for modification, and the semantics of the nonhead determines which of the slots fit. In the case of (21a, c), mill opens slots for, among others, the source of energy and the output of the operation. Wind fits the former slot and flour the latter. Of course, there are other slots, as in town mill (‘in the service of’) or riverside mill (‘located at’). Also the contrast in (27) can be accommodated if it is assumed that fertility is desirable and headache is not. However, without a proper theory of slots, the Variable R Condition does not offer any explanations.

For a relatively long period after 1978, the interpretation of compounds was low on the agenda of morphological research. Presumably, this is because the question was associated with generative semantics, and the predominant currents of generative grammar focused on the generation of grammatically correct forms. It is only at the start of the 21st century that some new proposals have been made. Three such proposals are compared by ten Hacken (2016). Two of them can be seen as building on the insights of Levi (1978) and Allen (1978) as well as on the critical reception of their ideas.

Lieber (2004) presents a theory of word formation that includes a mechanism for the coindexation of slots. It can be interpreted as providing the principled approach to slots that is missing in Allen’s (1978) proposal. In Lieber’s theory, the meaning of lexical entries consists of a skeleton and a body. The skeleton is a formalized part of the meaning that is expressed by a closed class of features. The body encodes the remaining aspects of meaning, some of which may be formalized as features, others only stated informally. Lieber (2016, p. 40) gives the example of cat, where [+material ([ ])] is a feature of the skeleton, <+animate> a feature of the formalized part of the body, and ‘meows’ belongs to the informal, more encyclopedic part of the body. As an example of a case where the coindexation expresses contrastive readings, Lieber (2016, p. 45) gives (28).

(28)

Compounding in Morphology

The contrast in (28) involves verbal compounds. The transitive verb celebrate has two slots, and the contrastive interpretation of the roles of family and birthday in (28) can be correlated with the interpretation of the nonhead as going into the first slot, corresponding to the subject, in (28a), but into the second slot, corresponding to the object, in (28b).

Compounds of the type illustrated in (21) and (27) do not have the same possibility of contrastive coindexation. However, in all cases, the precise interpretation of what coindexation means uses information from the body. The idea that a family is a group of people and as such <+sentient>, as opposed to birthday, qualifies it to be in the first slot of celebrate and at the same time indicates the nature of the event. In the case of (27), pill will have the encyclopedic information that it is taken with the purpose of changing something in the physical body of the person taking it. The encyclopedic information that fertility is desirable and a headache is not can then be used to make sense of the coindexation for the two compounds in (27).

Jackendoff (2009) presents a system for the interpretation of compounds in the framework of his Parallel Architecture (PA; cf. Jackendoff, 2002). Somewhat modified versions of this system are presented in Jackendoff (2010, pp. 413–451; 2016). The most striking aspect of his system is a set of ‘basic predicates.’ The number and characterization of these predicates vary slightly in the different versions, but several of them correspond to Levi’s (1978) RDPs, for example, cause and make. The crucial difference between Jackendoff’s basic predicates and Levi’s RDPs, however, is that the basic predicates are embedded in a much richer system for expressing the meaning of compounds. In particular, the combination of various mechanisms in a generative way results in an in principle unbounded number of possible relations. For barbershop, Jackendoff (2016, p. 34) gives the analysis in (29).

(29)

Compounding in Morphology

In (29a), it is formalized that shop refers to a place which has the proper function (pf) that someone sells something in it. In (29b), it is formalized that barber designates a person whose occupation (occ) is to cut hair. In the compound barbershop, the representation of barber is inserted into the representation of shop, but only after an operation of reprofiling. As represented in (29c), this operation puts the person profiled in (29b) into the position of α‎ with which it was coindexed. In (29d), most of the meaning is ultimately based on the lexical entries of the components. This is not because the components have a verbal root, but because for primary nouns as well, the semantic component of PA gives lexical conceptual structures with a great deal of relevant detail.

In all the approaches seen so far in this section, the interpretation of compounds is based on two sources of information. On one hand, there are the rules of compounding, contributing, for instance, RDPs or basic functions. On the other, there is the semantics of the components. The latter is the main source of information in Lieber’s system, but also very important in Jackendoff’s. Lieber’s approach starts from the form and works toward the meaning. Jackendoff’s approach starts from the components and works toward the compound. As noted by Downing (1977, pp. 822–824), compounds have the function of providing a name for a concept. Therefore, a further source of information that determines the meaning of a compound is the concept to be named. In general, compounds are not formed in order to combine two components but in order to name a new concept. They are neologisms. Ten Hacken (2013b) argues for a separate word formation component in PA on the basis of this argument.

The onomasiological approach to word formation as developed by Štekauer (1998, 2005) takes the naming function of word formation as its central characteristic. Although the resulting system does not identify compounding as a separate phenomenon, Štekauer (2016) explores how different types of compounding are covered in it. The point of departure is that the system provides a series of ever more specific decisions leading from the identification of a concept to be named to the selection of a name for this concept. A central role in this process is taken by the so-called onomasiological types (OTs). Štekauer (2016, pp. 59–61) presents eight different OTs, which lead to the morphological expression of different semantic aspects. The type of difference giving rise to different OTs can be illustrated on the basis of the examples in (30).

(30)

Compounding in Morphology

In (30a), three components are expressed, an object (guitar), an action (play), and an agent (‑er). It illustrates Štekauer’s OT1, which expresses all components of the onomasiological structure. Other OTs leave one or two components unexpressed or express two components in one morpheme. In (30b), the object is not expressed, which makes it an example of OT2. In (30c), the action is not expressed, as is typical of OT3. In (30d), representing OT5, the action and the agent are jointly expressed by cheat. It should be noted that Štekauer’s theory is much richer than only an expression of the contrasts in (30). The OTs illustrated here can also be applied to other types of word formation. Thus, writer also belongs to OT2 and novelist to OT3. The distinction between derivation and compounding is orthogonal to the distinction between OTs.

An important difference between the onomasiological approach and the other approaches seen in this section is that by starting from the concept, the ambiguity of a compound is not relevant. Starting from the concept ‘a mechanism for grinding grain into flour’ and working gradually toward the compound flour mill, there is no point at which the possibility of interpreting flour mill in a way parallel to windmill comes up.

In general, it can be said that there are three sources of information that can be used in the interpretation of compounds. One is the lexical entries for the elements that are combined. Then there is the rule for combining them, which may be equipped with specific conditions. Finally, there is the concept to be named. Different approaches have emphasized one or two at the expense of the others.

5. Correlations between the Different Issues

The distinction between delimitation, classification, formation, and interpretation is useful for expository reasons, but it is to some extent artificial. This is visible in the appearance of the same theories in different sections. After the presentation of the individual issues, it is therefore worth indicating more systematically how the four issues relate to each other.

There is a clearly observable opposition between two pairs of strongly related issues. Delimitation and classification can be called expository issues, whereas formation and interpretation are substantive issues. Expository issues are more descriptively oriented and originate in traditional grammar. Definitions of compounding and divisions into subtypes are used to come up with a systematic description. Substantive issues have an explanatory orientation and originate in theoretical approaches to language. A theory of language cannot dispense with an account of the formation and interpretation of compounds, but such an account does not have to identify a precisely delimited category of compounds. An example is Štekauer’s (2016) onomasiological theory, which treats compounds without identifying compounding as a class.

Historically, classification is the oldest issue, as it was already part of the Sanskrit grammarians’ concerns. In traditional grammar, the question of delimitation is less prominent in the sense that it follows from the classification. It only becomes an issue of its own when a crosslinguistic perspective is taken. Both issues can be seen also as primarily of a terminological nature. The question is, then, whether and how to define terms such as compound or subtypes, for example, exocentric compound.

The interest in the substantive issues only arose when explanatory theories became more important. As the discussion in sections 3 and 4 illustrates, the emergence of generative grammar played a central role in this process. Historically, three phases can be distinguished. The first was marked by the account of the formation and interpretation of compounds by means of a single, nondifferentiated mechanism. Exponents of this phase are Lees (1960) and Levi (1978). In the second phase, the emphasis was on formation mechanisms only. Examples include Allen (1978) and Selkirk (1982). Downing (1977) starts by criticizing the theories of the first phase, but instead of limiting herself to formation mechanisms, she makes some lucid observations that foreshadow the issues that emerge in the third phase. This third phase reintegrates the issue of interpretation into the theory, but separates it from the question of how compounds are formed. Here, Lieber (2004) and Jackendoff (2009) are good examples.

From a theoretical perspective, the substantive issues are more important than the expository issues. The extent to which definitions of terms designating (classes of) compounds are necessary is determined by the need to make and test theoretical claims depending on them. Whether a definition of synthetic compound, for instance, is needed depends on whether synthetic compounding is covered by a separate mechanism. When it is decided for what reason synthetic compound needs a definition this reason gives at least a clue as to the criteria this definition should be based on. In this way, the expository issues have become subordinate to the substantive issues.

At the same time, classifications have benefited from theoretical insights. An obvious example is Marchand’s (1969) use of Lees’s (1960) theory. However, it should be noted that a purely classificational use of a theory does not respect the theory’s original purpose. This was a source of conflict between Marchand and Lees. Also for other theories, it is important to see that they are not meant as a source for classification or delimitation, but as an explanatory account. Levi’s (1978) RDPs are not intended as a classification. When Levi (1978, pp. 280–284) gives lists of compounds for each RDP, her purpose is to illustrate RDPs, not to classify compounds. Jackendoff’s (2009) basic functions are even less suitable for a classification, because they are meant as only one component of a much more complex mechanism for the formation and interpretation of compounds. Similarly, for Štekauer’s (2016) OTs, it is not a problem when for some compounds it is hard to determine which OT they belong to. As long as all compounds can be accounted for, the purpose of the theory is fulfilled. As long as each OT has cases that are not covered by any other OT, its separate existence is justified.

The development of the research in compounding has seen a gradual shift of emphasis. Whereas expository issues were the focus of attention until the mid-20th century, substantive issues have become more central since then. Within the substantive issues, the interpretation has gained in importance in the last decades, after being in the background in the final part of the 20th century. Although no agreement has been reached in many of these issues, insight has grown.

Further Reading

Edited Volumes

Libben, G., & Jarema, G. (Eds.). (2006). The representation and processing of compound words. Oxford: Oxford University Press.Find this resource:

Lieber, R., & Štekauer, P. (Eds.). (2009). The Oxford handbook of compounding. Oxford: Oxford University Press.Find this resource:

Scalise, S., & Vogel, I. (Eds.). (2010), Cross-disciplinary issues in compounding. Amsterdam: Benjamins.Find this resource:

ten Hacken, P. (Ed.). (2016). The semantics of compounding. Cambridge: Cambridge University Press.Find this resource:

Special Issues of Journals

Scalise, S. (Ed.). (1992). The morphology of compounding, Rivista di Linguistica, 4(1).Find this resource:

Gaeta, L., & Grossmann, M. (Eds.). (2009). Compounds between syntax and the lexicon, Rivista di Linguistica, 21(1).Find this resource:

Bisetto, A. (Ed.). (2009). Compounds crosslinguistically, Lingue e linguaggio, 8(2).Find this resource:

Szpakowicz, S., Bond, F., Nakov, P., & Kim, S. N. (Eds.). (2013). On the semantics of noun compounds, Natural Language Engineering, 19(3).Find this resource:

Monographs

Lees, R. B. (1960). The grammar of English nominalizations. Bloomington: Indiana University Press.Find this resource:

Levi, J. N. (1978). The syntax and semantics of complex nominals. New York: Academic Press.Find this resource:

Selkirk, E. O. (1982). The syntax of words. Cambridge, MA: MIT Press.Find this resource:

ten Hacken, P. (1994). Defining morphology: A principled approach to determining the boundaries of compounding, derivation, and inflection. Hildesheim: Olms.Find this resource:

Lieber, R. (2004). Morphology and lexical semantics. Cambridge: Cambridge University Press.Find this resource:

Booij, G. (2010). Construction morphology. Oxford: Oxford University Press.Find this resource:

Language-specific handbooks

Marchand, H. (1969). The categories and types of present-day English word-formation: A synchronic-diachronic approach (2d ed.). München: Beck.Find this resource:

Bauer, L., Lieber, R., & Plag, I. (2013). The Oxford reference guide to English morphology. Oxford: Oxford University Press.Find this resource:

Fleischer, W., & Barz, I. (2012). Wortbildung der deutschen Gegenwartsprache (4th ed.). Berlin: De Gruyter.Find this resource:

de Haas, W., & Trommelen, M. (1993). Morfologisch handboek van het Nederlands: Een overzicht van de woordvorming. ’s‑Gravenhage: SDU.Find this resource:

Moyna, M. I. (2011). Compound words in Spanish: Theory and history. Amsterdam: Benjamins.Find this resource:

Ralli, A. (2013). Compounding in Modern Greek. Dordrecht: Springer.Find this resource:

References

Allen, M. R. (1978). Morphological investigations (Unpublished PhD diss.). University of Connecticut, Storrs.Find this resource:

Bauer, L. (1998). Is there a class of neoclassical compounds and if so is it productive? Linguistics, 36, 403–421.Find this resource:

Bauer, L. (2008). Dvandva. Word Structure, 1, 1–20.Find this resource:

Bloomfield, L. (1933). Language. New York: H. Holt.Find this resource:

Booij, G. (2010). Construction morphology. Oxford: Oxford University Press.Find this resource:

Dardano, M., & Trifone, P. (1985). La lingua Italiana: Una grammatica completa e rigorosa. Bologna: Zanichelli.Find this resource:

Di Sciullo, A. M., & Williams, E. (1987). On the definition of word. Cambridge, MA: MIT Press.Find this resource:

Downing, P. A. (1977). On the creation and use of English compound nouns. Language, 53, 810–841.Find this resource:

Dressler, W. U. (2006). Compound types. In G. Libben & G. Jarema (Eds.), The representation and processing of compound words (pp. 23–44). Oxford: Oxford University Press.Find this resource:

Fleischer, W. (1969). Wortbildung der deutschen Gegenwart­sprache. Leipzig: VEB Bibliographisches Institut.Find this resource:

Giegerich, H. J. (2009). The English compound stress myth. Word Structure, 2, 1–17.Find this resource:

de Haas, W., & Trommelen, M. (1993). Morfologisch handboek van het Nederlands: Een overzicht van de woordvorming. ’s‑Gravenhage: SDU.Find this resource:

ten Hacken, P. (1994). Defining morphology: A principled approach to determining the boundaries of compounding, derivation, and inflection. Hildesheim: Olms.Find this resource:

ten Hacken, P. (2003). Phrasal elements as parts of words. In E. Hajičová, A. Kotěšovcová, & J. Mírovský (Eds.), Proceedings of CIL17. CD-ROM, Praha: Matfyzpress, MFF UK.Find this resource:

ten Hacken, P. (2009). Early generative approaches. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of compounding (pp. 54–77). Oxford: Oxford University Press.Find this resource:

ten Hacken, P. (2010). Synthetic and exocentric compounds in a parallel architecture. Linguistische Berichte Sonderheft, 17, 233–251.Find this resource:

ten Hacken, P. (2012). Neoclassical word formation in English and the organization of the lexicon. In Z. Gavriilidou, A. Efthymiou, E. Thomadaki, & P. Kambakis-Vougiouklis (Eds.), Selected papers of the 10th International Conference of Greek Linguistics (pp. 78–88). Komotini: Democritus University of Thrace.Find this resource:

ten Hacken, P. (2013a). Compounds in English, in French, in Polish, and in general. SKASE Journal of Theoretical Linguistics, 10, 97–113.Find this resource:

ten Hacken, P. (2013b). Semiproductivity and the place of word formation in grammar. In P. ten Hacken & C. Thomas (Eds.), The semantics of word formation and lexicalization (pp. 28–44). Edinburgh: Edinburgh University Press.Find this resource:

ten Hacken, P. (Ed.), (2016). The semantics of compounding. Cambridge: Cambridge University Press.Find this resource:

ten Hacken, P. (2016). Three analyses of compounding: A comparison. In P. ten Hacken (Ed.), The semantics of compounding (pp. 211–232). Cambridge: Cambridge University Press.Find this resource:

Haeseryn, W., Romijn, K., Geerts, G., de Rooy, J., & van den Toorn, M. C. (1997). Algemene nederlandse spraakkunst. Groningen: Nijhoff.Find this resource:

Halle, M., & Marantz, A. (1993). Distributed morphology and the pieces of inflection. In K. Hale & S. J. Keyser (Eds.), The view from Building 20: Essays in linguistics in honor of Sylvain Bromberger (pp. 111–176). Cambridge MA: MIT Press.Find this resource:

Harley, H. (2009). Compounding in distributed morphology. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of compounding (pp. 129–144). Oxford: Oxford University Press.Find this resource:

Jackendoff, R. (2002). Foundations of language: Brain, meaning, grammar, evolution. Oxford: Oxford University Press.Find this resource:

Jackendoff, R. (2009). Compounding in the parallel architecture and conceptual semantics. In R. Lieber & P. Štekauer (Eds.), The Oxford handbook of compounding (pp. 105–128). Oxford: Oxford University Press.Find this resource:

Jackendoff, R. (2010). Meaning and the lexicon: The parallel architecture, 1975–2010. Oxford: Oxford University Press.Find this resource:

Jackendoff, R. (2016). English noun-noun compounds in conceptual semantics. In P. ten Hacken (Ed.), The semantics of compounding (pp. 15–37). Cambridge: Cambridge University Press.Find this resource:

Jespersen, O. (1942). A Modern English grammar on historical principles; Part VI: Morphology. Copenhagen: Munksgaard.Find this resource:

Koliopoulou, M. (2014). How close to syntax are compounds? Evidence from the linking element in German and Modern Greek compounds. Rivista di Linguistica, 26(2), 51–70.Find this resource:

Koliopoulou, M. (2015). Possessive/bahuvrīhi compounds in German: An analysis based on comparable compounds in Modern Greek. Languages in Contrast, 15, 81–101.Find this resource:

Konecny, C., & Autelli, E. (2015). Scale ripide, mobili e a chiocciolaSteile, Roll- und Wendeltreppen: Diverse categorie di combinazioni lessicali italiane del tipo “N + Agg o SPrep” e i loro equivalenti tedeschi – un confronto interlinguistico. In E. Lavric & W. Pöckl (Eds.), Comparatio delectat II: Akten der VII. Internationalen Arbeitstagung zum romanisch-deutschen und innerromanischen Sprachvergleich, Innsbruck, 6.–8. September 2012, Teil 2 (pp. 567–586). Frankfurt am Main: Lang.Find this resource:

Lees, R. B. (1960). The grammar of English nominalizations. Bloomington: Indiana University Press.Find this resource:

Levi, J. N. (1978). The syntax and semantics of complex nominals. New York: Academic Press.Find this resource:

Lieber, R., & Štekauer, P. (Eds.). (2009). The Oxford handbook of compounding. Oxford: Oxford University Press.Find this resource:

Lieber, R. (2004). Morphology and lexical semantics. Cambridge: Cambridge University Press.Find this resource:

Lieber, R. (2016). Compounding in the lexical semantic framework. In ten Hacken (Ed.), The semantics of compounding (pp. 38–53). Cambridge: Cambridge University Press.Find this resource:

Marchand, H. (1969). The categories and types of present-day English word-formation: A synchronic-diachronic approach (2d ed.). München: Beck.Find this resource:

Matthews, P. H. (1974). Morphology: An introduction to the theory of word structure. Cambridge: Cambridge University Press.Find this resource:

Panocová, R. (2015). Categories of word formation and borrowing: An onomasiological account of neoclassical formations. Newcastle: Cambridge Scholars Publishing.Find this resource:

Postal, Paul M. (1969). Anaphoric islands. In R. I. Binnick, A. Davison, G. M. Green, & J. L. Morgan (Eds.), Papers from the fifth regional meeting of the Chicago Linguistic Society (pp. 205–239). Chicago: University of Chicago.Find this resource:

Ralli, A. (2013). Compounding in Modern Greek. Dordrecht: Springer.Find this resource:

Roeper, T., & Siegel, M. (1978). A lexical transformation for verbal compounds. Linguistic Inquiry, 9, 199–260.Find this resource:

de Saussure, F. (1916). Cours de linguistique générale. Translated and edited by C. Bally & A. Sechehaye. Paris: Payot, 1981.Find this resource:

Scalise, S. (1992). Compounding in Italian. Rivista di Linguistica, 4(1), 175–199.Find this resource:

Scalise, S., & Bisetto, A. (2009). The classification of compounds. In Lieber and Štekauer (Eds.), The Oxford handbook of compounding (pp. 34–53). Oxford: Oxford University Press.Find this resource:

Selkirk, E. O. (1982). The syntax of words. Cambridge, MA: MIT Press.Find this resource:

Spencer, A. (1991). Morphological theory: An introduction to word structure in generative grammar. Oxford: Blackwell.Find this resource:

Štekauer, P. (1998). An onomasiological theory of English word-formation. Amsterdam, The Netherlands: Benjamins.Find this resource:

Štekauer, P. (2005). Onomasiological approach to word-formation. In P. Štekauer & R. Lieber (Eds.), Handbook of word-formation (pp. 207–232). Dordrecht: Springer.Find this resource:

Štekauer, P. (2016). Compounding from an onomasiological perspective. In P. ten Hacken (Ed.), The semantics of compounding (pp. 54–68). Cambridge: Cambridge University Press.Find this resource:

Szymanek, B. (2010). A panorama of Polish word formation. Lublin: Wydawnictwo KUL.Find this resource:

Trips, C. (2016). An analysis of phrasal compounds in the model of parallel architecture. In P. ten Hacken (Ed.), The semantics of compounding (pp. 153–177). Cambridge: Cambridge University Press.Find this resource:

Ward, G., Sproat, R., & McKoon, G. (1991). A pragmatic analysis of so-called anaphoric islands. Language, 67, 439–474.Find this resource:

Whitney, W. D. (1879). A Sanskrit grammar, including both the classical language, and the older dialects, of Veda and Brahmana. Leipzig: Breitkopf & Härtel.Find this resource:

Wiese, R. (1996). Phrasal compounds and the theory of word syntax. Linguistic Inquiry, 27, 183–193.Find this resource: