Wednesday, 1 May 2013

A new verb: task

In the British National Corpus, not a single use of task is tagged as a verb.
In the ukWaC (not pronounced you quack) it occurs 3,952 times, that is 2.5 times per million words. As this screen shot from the Sketch Engine shows, it has a strong preference for the passive - the B in the middle of the tag indicates a form of the verb to BE. In third position, the D in VBD indicates past tense and the N in VBN indicates past participle. Since past participles are only used after the auxiliaries HAVE and BE, we can expect a significant use of perfect aspect and passive voice. This is in fact the case.

In the third column we notice by, with and to, each one telling its own story:

  • with + noun phrase, most of which are Noun + of + Noun. For example: She will also be tasked with the development of strategic programming ...
  • with + -ing form, mostly followed by the object of the -ing verb. For example, The review team were tasked with identifying creative but practical actions to build on existing practice.
  • to + infinitive, mostly preceded by BE. The most frequent verbs: investigate, search, line, assist, and develop, produce, work. The -ing forms of some of these verbs can be seen in the screen shot also. For example, The crew were tasked to search the vehicle and establish that no persons were trapped inside.
  • by + the passive's agent. The data shows that the agent is never an individual, rather an organisation in authority. For example, Banks have been tasked by the Financial Services Authority to tackle fears over internet security ...

The delexical form, give ... task occurs in the BNC 1.2 per million and in ukWaC 1.1 per million. The search is give + 3 + task. It is mostly give someone the task of doing something. The adjectives between give and task in both corpora are quite similar. In both corpora we also assign tasks, a more formal verb.

Thus the modern construction, to be tasked, has assumed a special role in the domains of business, administration, politics, whereas give a task remains in use in general English.

Sunday, 28 October 2012

There would be a lot less violence in the world if ...

... if all animals were herbivores.

Zebras are herbivores.

Less than two weeks after posting this, the BBC Nature section reports research into why zebras evolved their stripes. The article is here.

From a linguistic point of view and one that represents the real world, it seemed odd to me that [animate creature] evolves [feature]. There are many transitive uses of this verb. The most frequent objects of the verb, according to the BNC, are shown in this screen shot from the Sketch Engine. These items clearly belong to a lexical set we might entitle approaches, thus
[person/organisation] evolves [approach].

Most other transitive uses of evolve have representatives of the animal kingdom as their subject, the implication being that the speakers/writers accept that the
creatures manipulate their environment over a long period to acquire the feature they now have. Another example:
Several groups of them specialise in this diet and each has evolved a long sticky tongue entirely independently...
However, the verb is far more frequently used intransitively and in perfect aspect, e.g.,
The character of the British countryside has evolved due to changing agricultural, industrial and recreational pressures ...
 I would say that this is a good example of the verb used as an ergative verb. See Wikipedia and MAELT on this topic.

Friday, 26 October 2012

Patterns of typical usage in "get over with"

This is a separable phrasal verb. Of its hundreds of appearances in the New Model Corpus (NMC), it occurs seven times without something in the middle, an object. Six of these are spoken language, such as:
  • How do I get over with you now ? Get over with me ? 

Not even seven swallows make a summer, so let’s move on to the patterns of typical usage. Using this CQL in the NMC
  • [lemma = "get"] []{1,7} [word = "over"] [word = "with"]

there are 216 occurrences. The {1,7} tells the concordancer to allow between 1 and 7 words to appear in between the other components. Get it over with has 112 of these, and get this over with 44. And 48 of all occurrences are preceded by let’s. These are patterns of typical usage which learners deserve to have highlighted.

This Wordle was created from the list of node forms that the Sketch Engine generates, minus the lemma GET. It is clear from not only the negative lexical words, e.g. agony, worst, fever, but also the attitudinal words e.g. antsy, fuck, shit that the phrasal verb is a vehicle for expressing negativity. This is despite the lack of inherent negativity in get and over and with.

Among the collocates on either side of it, quick, quickly, soon and hurry are significant. So are wish want sorry and please. 

Through our multiple exposures to this phrasal verb, with its negative objects, the need for speed and a tinge of imploring, we wouldn’t think to use this verb in a jolly way. So, even if the object were a happy event, such as a picnic, wedding, award ceremony, it is smeared by its environment and these events would not be perceived positively here. Due to the lack of attested examples, let us imagine a scenario in which someone might plausibly say, let's get this picnic over with. 

This is semantic prosody – through exposure we are primed to understand the whole meaning, and through using the phrasal verb in its pattern of typical usage, we reinforce its priming in ourselves and our interlocutors.

For our students to have active use of word, they need to know its patterns of typical usage.

Wednesday, 12 September 2012

New Presentations uploaded

Just to let you know ... for your information ... to whom it may concern:

In the subsection of my Methodologies and Approaches in ELT website called Data Driven Learning, I have uploaded the two presentations I gave during the summer:

Feedback more than welcome.

James Thomas

Saturday, 8 September 2012

Mixed marriages

English, as we all know, has both Germanic and Romance language roots. To quote two topic sentences from a Wikipedia page:

  1. All Germanic languages are satellite-framed languages.
  2. On the other hand, all Romance languages are verb-framed. 
In satellite-framed languages, verbs of motion contain require particles to express direction and how the verb is performed, whereas in verb-framed languages, such meanings are encoded in the verb itself. 

This partly accounts for the divide between Germanic and Romance vocabulary in English being multi-word and single word respectively.

Wednesday, 5 September 2012

Blame the tagger

Kennel is a verb?

In the UK WaC corpus, the phrase,  hunt kennels is tagged as N + V, which serendipitously turned up in the wordsketch of “hunt”. Kennel is tagged as a verb 500 times. The 170 kennels are mostly NNS (plural noun), i.e. wrongly tagged. Ditto the 96 kennelling, which are mostly gerunds, sometimes present participles as adjectives.  However, the 94 kennelled are mostly verbs, and mostly passive but sometimes simple past where they have a normal animate subject: someone kennelled their type of dog, pack.

This hunt set out to discover if kennel is used as a verb, and to defer to Rosemary Moon’s comment, you can verb any noun. Wordnet no doubt has a list of house nouns, many of which are as likely to be as verbable as kennel. To house is hardly an uncommon verb.

The odd thing about hunt kennels being tagged as N + V is that kennelling is not performed by inanimate subjects. And the tagger should take this into consideration. The unimaginative morphology of English that sees 's' as the third person singular ending and the third person plural ending is clearly at the heart of this problem. 

The tagger is also an inanimate non-sentient robot, so the buck stops with the programmers of taggers who are both animate and sentient beings.

Tuesday, 31 January 2012

Typical language vs cliché

On two recent occasions when I have presented the idea of vocabulary frames (see below) as a path to non-native speakers using vocabulary in more typical ways than they do when they work from an L1 starting point, one person in each setting has politely insinuated that such an approach must produce very clichaic language.

It seems that such people don't realise that language is full of prefabricated chunks. And that's what makes speech and writing sound "natural". What I learn from this is the need to start with different assumptions when introducing the value of prefab language. Vocabulary frames being but one type.

An example of a vocabulary frame is:
X takes priority over Y
But can any noun phrase occupy the X and Y positions?
Up to a point.
People and things, both concrete and abstract, can by X. And Y is limited to things which X would logically take priority over.
So thse are relatively open semantic fields when compared with the next example.

In X regales Y with Z
  • regalers are unlikely to be trees
  • the recipient is unlikely to be a TV 
  • the gift is unlikely to be a window
The semantic category of these three positions is restricted, not by grammar rules, but by semantic preferences. If Z is chocolates, then X is likely to be male and Y female. But if Z is the far more likely stories/tales/adventures, Y is likely to be a group.

The Macmillan dictionary is on the right track with its entry for regale. It says what the word is and does, but does it indicate what we do not do with the word? This is great source of error in non-native language.

Likely = tendency, i.e., that which is probable.

In the slot and filler approach to grammar, the tree regaled the TV with a window is a perfectly acceptable sentence. No syntactic rules are violated. How sad is that!

Wednesday, 3 June 2009

A use of articles

Is it more probable that we say a noun of a noun or the noun of a noun?

This pair of queries answers the probability question quite dramatically.

[word = "the"][tag = "N.."][word = "of"][word = "a"] 50,100 hits in the BNC
[word = "a"][tag = "N.."][word = "of"][word = "a"] 6,653 hits in the BN

An examination of the situations in which the a N of a N is used is an important step, since 6,653 is not a number of hits that can be dismissed in the same way we dismiss the 22 hits for seperate as opposed to 11,868 for separate.

These are rather bland queries, not permitting any other elements in the noun clauses.

Tectogrammatical Level Annotation

Tectogrammatical is not only a new word for me, but a new concept.