3.1. the non-linguistic cognitive literature

In non-linguistic cognitive research, e.g., Borst and Egelhaaf (Reference Borst and Egelhaaf1989) and Borst (Reference Borst2000), visual motion detection is seen as a delay-and-compare mechanism, i.e., a comparison of successive images with conflicting visual information;^{Footnote 5} see also Dellen and Wessel (Reference Dellen, Wessel and Squire2009). In the same vein, Dodge and Lakoff (Reference Dodge, Lakoff and Hampe2005) quote neuroscientific research revealing that the same brain structures are used when imagining or talking about an action and when actually executing the same action (pp. 73–74). These scholars also talk about the probability that processing path and manner of a movement may involve different parts of the brain:

Blaser and Sperling (Reference Blaser and Sperling2007) distinguish between perceptual (or visual) and conceptual motion. Perceptual motion is a direct “first-order” observable movement, “a mechanism that yields a conscious sensation of motion” (p. 627). Conceptual motion implies a different kind of processing, called a “cognitive simulation” (p. 627). Such processing is of a much more abstract nature and “has little restriction on the visual tokens that may be displaced” (p. 627). In a psychophysical experiment, the authors contrast luminance-based perceptual motion (a displaced pattern of luminance over time) with semantic-based conceptual motion (a shifting pattern of words and non-words). Based on the processing time of the observers, the authors conclude that whereas perceptual motion has “access to special-purpose motion mechanisms of the visual system” (p. 627), conceptual motion “would seem to be distinctly second-class” (p. 627) and dependent on the aforementioned cognitive simulation of perceptual motion.

3.2. durst-andersen and smith

Within the linguistic cognitive research, Durst-Andersen (1992, pp. 53–63, 2000, pp. 59–66) takes his starting point in “the perceptual notion of picture” (2000, p. 59) and argues that humans generally distinguish between two kinds of real-world phenomena or situations, both of which can be described in terms of the relationship between figure and ground:

1. a. States, which are perceived as a stable figure on a stable ground, providing a stable picture; ‘states’ are referred to by state verbs such as be, have, sit, lie, stand, etc.

2. b. Activities, which are perceived either as an unstable figure on a stable ground or as a stable figure on an unstable ground, providing an unstable picture; ‘activities’ are referred to by activity verbs such as carry, drive, walk, swim, crawl, etc.

Only these two (atelic) situation types can be perceived via direct observation, but in addition hereto, real-world experience has taught us that some activities can result in certain states, and some states can be the result of certain activities. This provides us with a third (telic) situation type:

1. c. Actions, which are mental constructs combining a particular activity with a particular state. When an ‘activity’ is conceived as part of an ‘action’, it is classified as a process, and the ‘state’ is classified as an event. ^{Footnote 6}Action verbs are verbs such as give, sell, buy, leave, kill, etc.

Smith (e.g., 2005a, 2005b, 2009) uses the observations of Durst-Andersen and Blaser and Sterling mentioned above to distinguish between simple and complex situations. ‘Simple situations’ include Figure, Ground and “simple Motion or Locatedness”, termed activities and states, respectively. ‘Complex situations’ consist of “observed or expected interdependencies between simple ones” (2005a, §2.2) and are labelled actions or relocations. These are mental constructs that relate ‘activities’ and ‘states’ to each other “not on a first-order perceptual level, but on a higher-order conceptual level” (§2.2.).

4.1. Germanic and Romance lexicalization of verbs

Regarding verb lexicalization, the distinction between a ‘picture’ and an ‘idea’ is a consistent extension of Talmy’s manner vs. path distinction. On the one hand, motion verbs containing the manner component denote a visibly specifiable activity that can be represented as a picture, as is seen in Figure 1.

Fig. 1. Representations of motion expressed by manner verbs.

On the other hand, path verbs convey an abstract idea of a movement from one location to another. Here, information is given about the relation between the figure that moves and the point of departure, traversal, or arrival, of the motion.^{Footnote 8} The way or manner in which the movement takes place is not rendered explicit but is so to say cognitively assumed or “simulated”, as Blaser and Sperling (2007, p. 627) put it, cf. section 3.1.

Blaser and Sperling’s description of ‘conceptual motion’ as a “cognitive simulation [which] has little restriction on the visual tokens that may be displaced” can be applied to the selectional restrictions of the Romance motion verbs in comparison with the Germanic ones. For instance, the Italian verb entrare ‘enter’ may select any kind of subject that can move or be moved, whereas Danish verbs denoting a movement-into-something are much more particular in their selection of the subject, as shown in Table 1.

table 1. Equivalent lexicalization of motion-into-something in Italian, Danish, and English

Unlike the Italian verb entrare, but like the English manner verbs (the first ones mentioned in the right-hand column of Table 1), all the Danish verbs require a subject that is congruous with the denoted manner. Thus, there is a strict correspondence, or congruence, between the manner component and the figure. Entities, animate as well as inanimate, must have specific physical features to be able to perform or undergo particular movements (as well as other actions or activities, as we shall see below). Thus, in such cases, the figure component could be defined as a resulting and secondary consequence rather than a direct and primary lexicalization such as the one found in English verbs like rain and spit (cf. Talmy, Reference Talmy2000, pp. 57–58).^{Footnote 9} Other cases of primary figure lexicalization are verbs such as (non-metaphorical) flow, stream, gush, spout, etc., which indicate that the entities involved are fluid.

As seen in the last three examples of Table 1, an Italian sentence may indicate the manner in an independent, typically adverbial constituent, often a prepositional group or a gerund, as in the cases cited here. Generally, however, such specification is only added if deemed particularly necessary, and it may be established elsewhere in the co- or context. On the other hand, the Danish and the English manner verbs cannot avoid expressing this specification, cf. the quote of Jakobson above.^{Footnote 10}

The right-hand column of Table 1 shows the ‘double set’ of English verbs denoting motion-into-something: one equivalent to the Danish verbs lexicalizing the manner component and one equivalent to the Italian lexicalization of path, ‘inherited’ from Latin and French. Such examples of mixed Germanic–Romance heritage have prompted scholars such as, e.g., Roberts (1939, p. 23) to define English as “the standard example of a hybrid language”.

4.2. what kind of manner?

Even with the specifications cited above, it may be argued that the component manner could cover more semantic grounds than those described by Talmy’s ‘self-contained motion’, and that manner verbs could be divided into different subtypes, as suggested, e.g., by Alonge (Reference Alonge and Agostiniani1997) and Korzen (Reference Korzen and Cresti2008). Where Talmy’s ‘self-contained motion’, as well as most of the verbs mentioned in the preceding subsection denote manner in terms of what could be defined as the ‘form’ of the movement (as depicted in Figure 1), a verb like sail listed in Table 1 denotes manner in terms of means of transportation. Such verbs, denoting the ‘means of motion’, are – just like the ‘form manner verbs’ mentioned in the preceding subsection – very common in the Germanic languages, especially in English, cf., e.g.:

1. (3) travel, drive, motor, fly, ski, skate, parachute, (bi)cycle, ride, sail/boat, bus, canoe, balloon, sleigh …

Of the verbs mentioned in (3), travel can be considered as a hyperonym merely indicating that a means of transportation has been used but not which sort.^{Footnote 11} Of these verbs, only travel, fly, skate, drive, (bi)cycle, ride, sail/boat have equivalent (non-phrasal) verbs in Danish:

1. (4) rejse, flyve, skøjte, køre, cykle, ride, sejle ^{Footnote 12}

Like the ‘form manner verbs’, verbs denoting manner in the sense of ‘means of motion’ are (more or less felicitously) depictable; see Figure 2.

Fig. 2. Representations of ‘means of motion verbs’.

It is characteristic that such verbs are very rare in the Romance languages:

1. (5) Italian: viaggiare ‘travel’, pedalare ‘bicycle’, sciare ‘ski’, pattinare ‘skate’, paracadutarsi ‘parachute’

2. French: voyager ‘travel’, patiner ‘skate’, se parachuter ‘parachute’

Yet another kind of manner is lexicalized in some motion verbs across the languages, such as the following Italian and English cases:

1. (6) accasciarsi/stramazzare ‘collapse, fall heavily’, fuggire/scappare ‘flee, escape’, sgattaiolare ‘sneak out’

Here, manner is not seen as form or means, but rather as the intensity of the movement, in accasciarsi/stramazzare describable with the adverb pesantemente ‘heavily’, in fuggire/scappare with the adverb velocemente ‘quickly’, and in sgattaiolare with the adverbs velocemente and/or silenziosamente ‘silently’. In these cases, and unlike verbs denoting form or means, the particular manner component is not representable as a picture, and, like the verbs cited in (1), the verbs in (6) are all unaccusative and telic.^{Footnote 13}

Thus, whereas the verbs in (3) and (4), illustrated in Figure 2, are very similar to other manner verbs, both syntactically and in their frequency in the two language families, the verbs in (6) should be analyzed as path verbs with an additional semantic component of intensity – and fuggire, scappare ‘flee, escape’ also with a component of purpose, i.e., ‘to get away (typically) from a dangerous situation’.

4.3. extensions of the motion verb lexicalization pattern

The distinction between concrete and relatively precise ‘picture verbs’, applicable to a limited number of entity categories and typical of the Germanic languages, and abstract and more generic ‘idea verbs’, applicable to a much larger number of categories and typical of the Romance languages, is not limited to motion verbs but can be extended to many other verb groups (albeit, obviously, not without exceptions). In this section, I cite a few Italian–Danish examples; many more French–Danish examples can be found in Herslund (1997, pp. 18–30).

In parallel to motion verbs we find – also in Talmy’s framework – position verbs, and similarly to the Italian motion verb andare, which just means ‘to move’, the Italian verb stare ‘to be’ can correspond to all five Danish/English position verbs, sidde ‘sit’, ligge ‘lie’, stå ‘stand’, knæle ‘kneel’, and hænge ‘hang’. Again, it is characteristic that the states denoted by these five verbs can all be represented as pictures, unlike stare; see Figure 3. As in the last three cases of Table 1, a specification may be added to the Italian verb: stare seduto ‘be sitting/seated’, stare sdraiato ‘lying down’, stare in piedi ‘on foot’, stare in ginocchio ‘on knee’, stare appeso ‘hanging’, but, as before, this will only happen if strictly necessary.^{Footnote 14}

Fig. 3. Representations of position expressed by manner verbs.

Also in other cases, Danish usually specifies the position of an entity. Table 2 shows how either the position or the shape of the object is decisive for the choice between the abovementioned position verbs in Danish. In all cases, the Italian verb, ‘V’, can be essere, stare, or trovarsi ‘be (located)’.^{Footnote 15}

table 2. Examples of position verbs in Italian and Danish

The difference between abstract and generic Italian verbs and concrete and precise Danish verbs is found in many other cases. Table 3 shows how the Italian verb ‘of destruction’ rompere, which, in most of its usages, according to Lo Zingarelli 2010 , has the synonyms spezzare and spaccare, corresponds to no less than fifteen different Danish verbs: brække, knække, bryde, gennembryde, afbryde, flække, knalde, knuse, mase, baldre, smadre, ødelægge, kløve, hugge, and brænde depending on the manner of the action and the figure of the object destroyed.^{Footnote 16} The Italian examples in Table 3 all come from Lo Zingarelli 2010.^{Footnote 17}

table 3. Italian and Danish ‘verbs of destruction’

Table 4 cites some Italian and Danish (transitive and intransitive) kitchen verbs. Again, the Italian verbs are more abstract and denote what are, from the perspective of Danish, different types of activities. On the other hand, the Danish verbs are more precise. The more precise Italian verbs in parentheses at the bottom of the left-hand column exist but generally belong to a special linguistic register and/or to special text genres such as cookbooks.

table 4. Italian and Danish kitchen verbs

And while we are in the kitchen region: if food (or something else) is evaluated positively, in Italian the praise would sound: è buono, che buono, che bello. But in Danish, it is generally necessary to specify the manner in which it is ‘good’; see Table 5.

table 5. Italian and Danish evaluation expressions

In all the cases cited in this section, the Italian verbs (and, in Table 5, adjectives) are more abstract and generic and may apply to many more action or activity types. On the other hand, the Danish verbs are more precise and concrete; unlike the Italian verbs, they express the manner of the verb situation and – as a consequence – often also the kind or figure of the object that is involved in it.

4.4. Germanic and Romance lexicalization of nouns

Extending the distinction between precise and concrete lexemes (which can denote relatively few instances) and generic and abstract lexemes (which can denote a larger number of instances) to nouns, we find exactly the opposite picture in the Germanic and Romance lexicalization patterns. Compared with Romance lexicalization of nouns, Germanic primary lexicalization (i.e., of basic lexemes, not compounds or derivatives) occurs at a hyperonymic level covering more entity categories, i.e., having a larger extension. To achieve the same precise denotation (i.e., larger intension and smaller extension) of a Romance primary lexicalization, a Germanic language will typically create a N+N compound, as is seen in Table 6.

table 6. Two examples of nominal lexicalization in Italian, Danish, and English

The basic Danish lexeme vogn can be used for all the mentioned subtypes and stol for the first four instances of ‘chairs’. In these cases, the specifying modifier (the first part of the compound) will only be added where necessary for the interpretation of the noun, which – as with the Romance verbs – may rely on other elements in the co- or context. On the other hand, the Italian lexemes in the left-hand column cannot avoid expressing this specification. As Table 6 shows, Italian does not have a hyperonym equivalent to stol. ^{Footnote 19}

Again, English appears as a mix of the Germanic system with N+N compounds (e.g., breakdown van, baby carriage, deck chair) and the Romance system with primary lexicalization at a hyponymic level (e.g., throne, pulpit, court).

The (primary) Germanic and Romance lexicalization of artefacts can be seen in the light of the four semantic components, or ‘explanatory categories’ (Müller, Reference Müller, Korzen and D’Achille2005, p. 65), that Pustejovsky (1995, pp. 76–104) includes in his ‘qualia structure’ of lexical items, viz. the constitutive, formal, telic, and agentivequalia. In Danish, it is the function of the object (corresponding to the telic quale) that is the dominant and distinctive feature of the basic lexemes, the feature which all subtypes have in common and which distinguishes them from other categories and subtypes; in Table 6, a vehicle used for transportation and a piece of furniture used to sit or stand on. In Italian, the lexicalization is based on external, visible aspects of the object (corresponding to the constitutive and formalqualia), and each subtype is morphemically unique. For more Italian and Danish examples, see, e.g., Korzen (Reference Korzen, Korzen and Marello2000, Reference Korzen and D’Achille2005b); for more French and Danish examples, see Herslund and Baron (Reference Herslund, Baron and Baron2003) and Herslund (1997, pp. 30–43); for Spanish examples, see Müller (Reference Müller, Korzen and D’Achille2005).^{Footnote 20}

4.5. the complete picture: endocentric and exocentric languages

Combining the above-mentioned tendencies for verb and noun lexicalization in Germanic and Romance languages, we obtain an interesting set of patterns. Typical Germanic lexicalization produces concrete and precise verbs, based on the external picture of the verb situation – including the figure of the entity involved – and abstract nouns based on the function of the artefact, i.e., on the idea of what purpose the object can serve. On the other hand, Romance lexicalization tends to produce abstract verbs, based on a generalized and/or conceptualized idea of the verb situation, and precise nouns based on the external aspects, the picture, of the object. This can be illustrated as in Table 7.

table 7. Typical lexicalization patterns of Germanic and Romance verbs and nouns

The more precise lexical content (i.e., larger intension) entailing a more limited extension of applicable instances thus lies in the Germanic verbs and in the Romance nouns, and, as stated in the ‘Introduction’, the conception of the verb as constituting the ‘centre’ of the proposition and the noun(s) as placed outside this centre has prompted the TYPOlex group to use the terms endocentric (i.e., Germanic) vs. exocentric (i.e., Romance) languages. In ‘endocentric languages’ the informative weight is concentrated in the verb, which also selects the (abstract) nominal arguments linked to it. In ‘exocentric languages’ the informative weight lies in the nouns, whereas the verbs that link the nouns together are relatively abstract and generic and do not put as heavy selectional restrictions on the arguments.^{Footnote 21}

4.6. motion verbs differing from these patterns

The lexicalization patterns shown in Table 7 describe important tendencies, but not, however, rules without exceptions. Although Talmy has a point in stating (with respect to the lexicalization of motion) that a language uses only one lexicalization type for the verb “in its most characteristic expression of Motion” (cf. Section 2), endocentric path verbs and exocentric manner verbs, i.e., exceptions to the rules mentioned, are not difficult to find. As already stated, English may be defined as a ‘hybrid language’ on this account, but also Danish, a more ‘genuinely endocentric’ language, has path verbs, albeit fewer; cf. Table 8 (which does not pretend to be complete).

table 8. Endocentric path verbs

Most of the English verbs are inherited from Latin and French; exceptions are go, come, fall, and rise. In some cases, e.g., come, pass, disappear, the path component is or can be deictic; cf. footnote 8.

French manner verbs are quite frequent and common, even if path verbs are more so, both in types and in tokens (Herslund & Baron, Reference Herslund, Baron and Baron2003; Schøsler, Reference Schøsler and Birkelund2010).^{Footnote 22} Most French path verbs are unaccusative, telic, and select the auxiliary verb être ‘be’, whereas the manner verbs are unergative, atelic, and use the auxiliary avoir ‘have’; cf. Table 9.

table 9. French motion verbs^{Footnote 26}

Nevertheless, there are differences also within the Romance languages. If the pattern illustrated in Table 9 could be defined as ‘typically’ Romance, the Italian situation is somewhat more complex. Again, path verbs are more common both in types and tokens, see Table 11 in Section 4.7, but regarding manner verbs, we find two groups: one is similar to the French manner verbs: unergative, atelic, selecting auxiliary avere ‘have’; see the right-hand column of Table 10 (which, like Tables 8 and 9, does not pretend to be complete). Another group consists of verbs that occur either with that same meaning and combination (i.e., unergative, atelic, expressing solely manner and selecting avere), or with an unaccusative, telic meaning, combining path and manner end selecting the auxiliary essere ‘be’ (see the left-hand column of Table 10). In this, the latter group is very similar to the equivalent Danish verbs (Herslund & Baron, Reference Herslund, Baron and Baron2003; Korzen, Reference Korzen, Bache, Holsting and Kjær Nissen2012).^{Footnote 25}

table 10. Italian motion verbs with the manner component

table 11. Italian motion verbs, types and tokens in Lessico di frequenza dell’italiano parlato

Like the manner verbs listed in Table 1, the verbs in Table 10 select specific subjects, and in cases like colare, gocciolare ‘drip’, fluire ‘flow, stream’, and sgorgare ‘gush, spout’ we should probably talk about a primary lexicalization of figure and not ‘only’ congruence. In non-metaphorical usages, the entities involved must be fluid. In the other cases, it would be more precise to say that the figure must be congruous with the denoted manner.

In their unaccusative, telic usages, these verbs thus follow a fourth lexicalization type, a type D, in addition to the three mentioned by Talmy (cf. Section 2):

1. D. motion + path + manner (+ figure)

where the component figure is ‘primarily’ lexicalized only in colare, gocciolare, fluire, sgorgare.

Referring to “a minor system” in Atsugewi, Talmy (2000, p. 63) mentions the rare case of verbs “expressing the ‘MOVE’ notion in isolation”. However, it should be noted that the two Italian verbs girare, circolare may occur with the sole meaning of muoversi ‘move’, in cases such as:

1. (7) Ho girato tutta Roma per trovarlo.

2. ‘I have been all over Rome to find it’.

3. Circolare, circolare! [said, e.g., by a traffic policeman]

4. ‘Keep moving! Clear the street!’

In such cases, these verbs do not convey a path meaning of ‘circulate, move around in circles’.^{Footnote 27}

4.7. conclusion: Italian and French as exocentric languages

Even if Italian does have a consistent number of both path and manner verbs, it is interesting to examine the differences in frequency. A count of all motion verbs in the spoken language corpus LIP: Lessico di frequenza dell’italiano parlato (De Mauro, Mancini, Vedovelli, & Voghera, Reference De Mauro, Mancini, Vedovelli and Voghera1993), consisting of 490,000 words, gave the results shown in Table 11.

Considering that three of the cited manner verbs, correre, saltare, volare, can also express path (cf. Table 10), it seems more than fair to conclude that Italian – like French (cf. Section 4.6) – should be regarded as an exocentric language with a quite dominant occurrence of the abstract path component in motion verbs. However, more concrete and precise manner verbs are by no means excluded.

Further corroboration of this claim is found in a study carried out by the IMAGACT project (cf. Moneglia & Panunzi, Reference Moneglia, Panunzi, Cresti and Korzen2010, and footnote 16), which compared the frequency of verbs with and without the component manner, in their terminology verbi di attività and verbi generali respectively, in a corpus of spoken Italian of about 1.3 million words. The scholars found that the ‘general’ verbs, those without the manner component, were about twice as frequent regarding types and almost three times as frequent regarding tokens.